This paper presents an interesting study demonstrating the value of geology maps in enhancing hydrological understanding. The authors have developed a reclassification method that transforms the original geology map into numerical metrics related to hydrology, and they illustrate the added value of more detailed, small-scale maps. The findings in this paper are valuable for more effectively utilizing geology maps to improve hydrological insights, making it worthy of publication. However, I would like to raise several major and minor concerns that should be addressed prior to publication.

Major Concerns

1. It is unclear why the analysis incorporating climate and landscape attributes is conducted. This study focuses primarily on the value of geology map details, and most of the results are related to the geology map analysis. The analysis using catchment attributes appears to contribute little to the main objective and conclusions of the study. The authors should consider explaining in greater detail how this part of the analysis connects to the study’s primary conclusions.

2. The authors seem to assume a priori that there is an inherent relationship between geology metrics and hydrological signatures, and that a map producing a higher correlation coefficient (rs) is automatically superior. Although the authors attribute this to "physical understanding," from a physical perspective, hydrological signatures are also influenced by climate and land use factors. More detailed information on climate, land use, soil, and topography would also be helpful for interpreting hydrological processes. I suggest that the authors clarify this issue, explain the mechanisms by which geology metrics affect hydrology, and adjust some statements to avoid presuming that geology metrics are the dominant influence.

3. The inherent or fundamental differences among the three maps should be summarized somewhere in Section 2.2. At first glance, the differences appear to be in the spatial range resolution, but this factor does not actually explain the differences observed in the correlation analyses. Clarifying these intrinsic differences would help readers better understand why the regional map performs better.

Minor Concerns

L188: Consider mentioning the five selected basins for detailed analysis at this point.

L236: Should “Five” be corrected to “Four”?

Section 3.4.3: The use of the geology map seems to be missing here.

L328: A statistical analysis is needed to determine whether the higher rs derived from the continental map compared to the global map is statistically significant. A similar analysis should be conducted elsewhere when describing the differences among the three maps.

Paragraph around L330: In the right part of Figure 3, many basins exhibit much lower rs values with the continental map compared to the global map. We cannot simply regard the higher continental rs in the left part as an “added value” while ignoring the lower values in the right part. This discrepancy reflects the divergence between the two maps, as also shown in Figure 2. The divergence between the maps should be analyzed carefully, rather than simply judging which map performs better based solely on a higher rs.

Paragraph around L480: The conclusion that the regional map provides the most stable correlations appears inappropriate. Among the three regional metrics, only one consistently produces the same sign, while the other two have one and two exceptions, respectively. However, similar patterns can be found in many other metrics.

Figure 8: Consider adding separating lines between the different groups to enhance the figure’s readability.

L483: The value “0.93” is not found in Figure 8.

This paper provides an interesting assessment of geological information on streamflow signatures. The study of this work at separate scales is potentially interesting and a refreshing perspective for large-sample hydrology studies. The study could be a useful addition to the hydrological literature, but before I can recommend publication, the following points need to be clarified:

• “Normal” (typical) large sample hydrology analyses study (statistics) relationships across the entire study domain at once. This paper takes a different approach by studying local (basin-scale) relationships across many basins at once. This is a valuable addition to the literature, but this contrast needs to be better highlighted, as it fundamentally changes the question that is asked (the current paper does not check continental scale dominance of catchment attributes) but the local scale dominance of factors (across an entire continent). This different approach inherently changes the answers one gets, but will also change the question that is asked, and as such this should be better contrasted to the existing literature. Right now this 
• In the introduction, the manuscript reflects on why landscape appears to have a small impact on hydrology according to LSH studies. The work states this may be due to several reasons but maybe overlooks (according to my gut feeling, not that I have no formal evidence) the most obvious reason: the landscape can have a very important role on hydrology but how important this role is depends on the diversity of climates considered. Most LSH hydrology studies are across tremendous climate gradients (e.g. the current continental-scale study) and this easily dominates the role of many (but not all) hydrological signatures. Sure, landscape will (strongly) modify how incoming precipitation is partitioned, but, if climate differences are big enough, it cannot override these climate effects. And if we are not yet super successful at normalizing for climate effects, effectively learning about landscape in LSH studies across large geographical domains becomes challenging. If LSH were conducted with many catchments in a similar climate setting, the role of landscape would become (relatively seen) much more important and very likely easier (but still challenging, due to reasons you also state) to identify. 
• The approach that is used already seems to partly acknowledge this, as at large scales, it checks for correlations at the scale of a basin, which reduces climate gradients. Explain this. But also explain thereby that the results you get are “dominant controls” on the scale that you study, and not at the continental scale when the large scales is studied. 
• In the discussion, the paper also reflects on this by comparing it to previous studies “(Addor et al., 2018; Beck et al., 2015; Kratzert et al., 2019; Kuentz et al., 2017) that found that climate is a stronger control on signatures than landscape. The current manuscript needs to be more precise in stating that “if the scale of the assessed correlations changes, the question changes” (and thus the answer). 
• The choice to use Spearman correlations is OK, but not very convincing. I understand that every method comes with its problems (and advantages) but it would be good if the reader gets more confidence that a method that not only considers these individual correlations (but tests for interaction of more catchment attributes) would come to broadly similar conclusions.
• The use of the “maximum |r_s|” seems somewhat odd when not all groups have an equal amount of catchment predictors. Is there a risk that groups that have more predictors are considered to be more dominant, not because they physically are, but because they are more numerous? (I do not expect this to be dominant, but maybe good to consider).
• Correlation of catchment attributes and signatures seem to be interpreted as “correlation = causation”, but that is very speculative. I understand that this often happens in LSH, but here the approach is rather with a scattergun approach: the work studies a very long list of correlations (without physical hypotheses how various factors shape individual catchment attributes) and then picks the strongest correlation at the catchment scale. This seems sensitive to spurious correlations. 
• FIg 3: I think it would help if this information was also shown in a different way. For example, scatter plots between geology global values and the other categories. In addition, it would be very helpful it is shown how strongly the best predictors of the different classes are correlated with another (again with scatter plots)
• Fig. 4: this comparison between the effects (or correlations) of global geology vs continental geology is useful, but it would benifit from shwong a but more than just the data points and 1:1 line. What are the average values of both (you can show the average X and Y coordiante, and this summarizes how well one predicts vs the other, and how well they overall predict). You can also show the correlation (coefficient) between the two data points to show how strongly related they are to another. Such things could greatly help the interpretation of these plots.