I would like to thank the authors for the thorough revision of the paper and the response to my comments. The paper has substantially improved, although still some issues are not completely clear and sufficiently substantiated (e.g. the non-stationarity issue, see specific comment 14). Furthermore, the English writing style can be further improved. Spelling, style and grammatical errors can still be found in the manuscript. My specific comments and technical corrections are given below.
1. P1, L11: What is the ‘theoretical distribution fit’? The fit of what? This is not clear and cannot be understood from the abstract alone.
2. P2, L1: Is only CO2 included in this scenario? Greenhouse gas scenario seems to be more appropriate here.
3. P2, L22-33: The structure of this part is not very clear. Several issues are discussed here and there is a sudden transition from the topic of uncertainty to bias correction methods in line 26. Please try to improve the structure of this part of the introduction.
4. P2, L28: “… are dealt with …”; in this study or in other studies?
5. P2, L33: “… which is the main focus of the present paper …”; the main focus has not been introduced yet and moreover, besides hydrological model parameter uncertainty two other sources of uncertainty are the focus of this paper.
6. P3, L9-12: The aim of the paper is still not very clear. The authors suggest that the influence of hydrological model uncertainty is assessed while only the influence of hydrological parameter uncertainty is assessed (and not for instance model structure). Furthermore, it seems the spread of extreme indices is related to the spread due to different climate models and the distribution fitting error (or theoretical distribution fit as in the abstract). However, while the main idea of the paper seems to be to compare the effects of three sources of uncertainty (parameters, climate models, distribution fit) on the uncertainty in hydrological extremes, the research aim suggests something different.
7. P3, L13-19: I appreciate the explanation of the authors regarding the choice of the uncertainty sources, but would like to advise them to mainly connect this explanation to the previous work of Osuch et al. (2016) and not for instance mentioning that uncertainty related to ‘geomorphology of the catchment’ is not taken into account. I.e. this latter source seems to be very specific (and arbitrary) and might even be slightly incorporated in the hydrological model parameter uncertainty.
8. P5, L3-4: How can these references refer to the RCP scenarios, which have been developed after 2010?
9. P5, L26-28: Although this statement seems to be logical and obvious, this might depend on the case study/ catchment area considered?
10. P6, L7-9: It is not clear to me whether the appropriateness of bias correction has been investigated for the study area used in this paper or for other catchments. In the former case, more evidence needs to be provided that the different cases with/ without bias correction are appropriate and in the latter case the authors need be make clear that we can assume the same ‘bias correction conditions’ for their study area as compared to other study areas. The application of the hydrological model with climate data with and without bias correction is described in section 3.7, but already in this section (3.2) it is decided to only use one set of data (with or without bias correction).
11. P6, L23: “… following those studies”; why can we assume the same sensitive parameters as in ‘those studies’? This is still not clear. Moreover, it is still hard to imagine that the same set of six sensitive parameters has been found for low and high flows.
12. P7, L5-6: How did you combine the low flow and high flow parameter sets? Or did you run the HBV model with each set separately? Then, how is the performance of the HBV model with low flow parameters for high flows (and vice versa)? When calibrated for low flows (high flows) does the model still give reasonable results for other flow conditions and does the model give the right results for the right reasons?
13. P8, L7: Do the authors mean ‘annual maximum and minimum daily flows’? Why do they use annual minimum daily flows as an indicator of low flows? Wouldn’t a minimum value averaged over multiple days be more appropriate (e.g. 7 or 10 days)?
14. P9, L9-11: It seems to be contradictory to have stationarity in extreme flow events while at the same time the authors conclude that there are substantial differences (increases and decreases) between extreme flows for the reference and future periods (see Figure 5 and section 4.3). How should we see this? Is the comparison between uncertainty ranges derived using the two approaches a fair comparison?
15. P9, L23-24: How did you combine the 10 000 MC samples for GEV with 20 000 parameter sets from GLUE? How arbitrary is the result of this uncertainty analysis? Did you repeat the uncertainty analyses and were the results in line with the first uncertainty assessment?
16. P10, L4-6: I would still recommend to plot the individual model results in this figure, seven model combinations/ points do not justify the use of a box plot. In the caption it should be ‘annual mean temperature’ instead of ‘annual maximum daily temperature’.
17. P10, L9-10: Do you have evidence that the annual precipitation sum is an important driver of low flows or is for instance a six-month sum much more important?
18. P10, L10-12: The increase of the temporal variability with time is not very clear from Figure 2.
19. P10, L26-27: What are the maximum logNSE values for the calibration and validation periods?
20. P11, L20-25: The changes in annual minimum flows are also relatively smaller than the changes in annual maximum flows (see Figure 5). This can be clearly seen from this figure since the range of quantile values for the reference period is approximately the same for low and high flows.
21. P12, L22: Which confidence interval is considered here, 95% or 90%? In the first case, the upper and lower boundaries should be 0.975 and 0.025 respectively.
22. P12, L28-31: How should we read Figure 7? Are the different colours overlapping or additive? And how should we interpret the different points?
1. P3, L18: “… hydrological model parameter uncertainty …” instead of “… hydrological model uncertainty …”.
2. P4, L7-8: “Study area …” instead of “Study areas …”.
3. P4, L12: “… the river is characterized …” instead of “… the river characterized …”.
4. P4, L14: “… areal precipitation …” instead of “… aerial precipitation …”.
5. P6, L26: “… to eliminate …” instead of “… to eliminated …”.
6. P7-9: The description of the GLUE methodology is spread over three sections (3.3, 3.4 and 3.7) Try to structure this in a better way.
7. P9, l29-P10, L2: Please try to harmonize the terminology, such as ‘variability’ vs. ‘patterns’ and ‘projections’ vs. ‘changes’.
8. P10, L14-20: This has already been described in the methods and can be moved/ integrated there.
9. P13, L21: Please clarify ‘medium range’.
10. P14, L23-26: Is this conclusion for the 1971-2100 time series?