the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Trends in hydroclimate extremes: How changes in winter conditions affect seasonal baseflow and storage
Abstract. Northern ecosystems experience rapid climatic change at a rate where average temperatures are increasing above global averages. Yet, for the boreal snow-dominated catchments that rely on the winter snow accumulation and spring melt for sustained stream flow across preceding seasons, much remains unknown about how catchment water storage and baseflow are affected. Here we used 40 years of data from the boreal Krycklan catchment, placed into a 130 climate record from a nearby location, to test how 27 extreme climate change indices have been affected, and how these, in turn, can explain seasonal low flows during the winter and summer. Our results show that while annual temperatures have increased by 2.2 °C over the last four decades, even more distinct seasonal impacts were detected as exemplified by eight extreme indices demonstrating that winters have become warmer with less precipitation. The analysis also showed that summers have become warmer shown by four significant increases in climate indices. Using the significant winter indices to predict winter baseflow and winter/summer indices to predict summer baseflow we found that the accumulated degree day below zero (AFDD<0) was the best predictor of winter minimum flow and AFDD<0 and Summer maximum temperature (MaxTmax) were the best predictor of summer minimum flow. Additional isotopic analysis of stream flow partitioning found an increasing contribution of winter rain/snow in stream runoff during winter over the last 22 years, as well as a decreased contribution to the preceding summer stream flow. These findings imply that warmer winters have affected water storage and runoff patterns in the boreal catchment which can have important feedback on terrestrial ecosystems, particularly on water availability in later parts of the growing season.
- Preprint
(1339 KB) - Metadata XML
-
Supplement
(932 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on hess-2024-337', Anonymous Referee #1, 12 Feb 2025
The article is overall well written and well structured. The results obtained are interesting. However, my comments focus on the following two points:
- In the flow trend analysis, the authors use the linear regression method. But, from a statistical point of view, this method is not very appropriate, because it does not consider the autocorrelation effect (short-term persistence). Furthermore, the method is only valid for data with a normal distribution. However, some analyzed data (daily rainfall) do not follow the normal law. It would be interesting if the authors applied the classic non-parametric Mann-Kendall type test which considers this short-term persistence effect.
- In their study, the authors do not consider the impacts of autumn rains which can influence, to varying degrees, base flow in winter in very cold temperate regions. The exclusion of the fall season in the study was not rigorously justified.
- It would be important to add the geographical map of the region studied.
Citation: https://doi.org/10.5194/hess-2024-337-RC1 -
RC2: 'Comment on hess-2024-337', Anonymous Referee #2, 12 Feb 2025
This manuscript examines long-term trends and variability in the seasonal and annual temperature, precipitation and runoff in the Svartberget catchment, within the Krycklan catchment, located in the northern Sweden’s boreal zone. This manuscript compliments many of the past studies done within this catchment, adding a much-needed analysis on the long-term effect of changes in temperature and precipitation on the runoff.
Overall, the manuscript is well written and well structured. I believe that the current manuscript requires additional analyses and information to be publishable as a journal article in Hydrology and Earth System Sciences. Provided below are a list of questions, comments, and suggestions towards an improved version of the manuscript.
General Comments:
1) Line 85: With the hypothesis that warmer winters will result in higher runoff during the winter, exhausting summer baseflow, how much more runoff is expected to occur during your defined winter period from snowmelt? If temperatures remain below freezing during the winter, then I suspect an increase in winter baseflow would be from the previous autumn season. Would the changes in the timing of snowmelt most likely affect the spring season thus creating more low flow events in the summer?
2) Line 126: Why was the autumn season not used in this study? Late autumn discharge, enhanced by autumn rainfall and early snowmelt sessions, will influence the total winter flow. An analysis of the autumn season should not be omitted in this study.
3) Are there any strong connections of climate oscillations (e.g. AO or NAO) to the climate variables or climate indices that could then influence runoff in this catchment?
4) In order to detect trends and significant changes, I am wondering why a more suitable Mann-Kendall test and Sen’s slope was not employed on the time series?
5) Any map that you could provide for the readers of the Krycklan and Svartberget catchment, including location of sampling sites, met tower(s), and hydrometric stations?
6) Have you considered using a change-point detection on the time series to detect any regime shifts?
Specific Comments:
1) Page 1, Title: In the manuscript, you focus on the other seasons, not just winter. I suggest changing the title to be reflect your analysis.
2) Line 105: What is the difference between the two different data sets? Was the shorter period data set only missing daily maximum and minimum temperatures? Why not use the 1982-2022 data set to detect long-term trends and “extreme climate change”?
3) Line 110: What is the difference between the two stations? Was any analysis conducted to see the similarities between the two locations that are 150 km apart?
4) Line 116: What percentage of the data set was gap-filled and the quality of the data? The reference, Karimi et al. (2022), does not mention this information for the 1982-2022 time period.
5) Line 144: Do you avoid possible inhomogeneity as described in Zhang et al. (2005)? Zhang, X., Hegerl, G., Zwiers, F.W. and Kenyon, J. (2005) Avoiding Inhomogeneity in Percentile-Based Indices of Temperature Extremes. Journal of Climate, v18, 1641-1651. DOI:10.1175/JCLI3366.1.
6) Line 144: For Table S1, could you provide more details on how these are calculated?
7) Line 146: For the maximum 1-day precipitation total, what if the precipitation event starts at night and ends in the morning? How do you account for this?
8) Line 175: Define SOI as the Seasonal Origin Index here.
9) Line 191: The change in average daily temperatures from two different years is misleading and is best to just stick to the slope value of 2.2. Similar comment for the select years on Page 8, unless you are mentioning the upper and lower boundary of the variability.
10) Figure 1B: What does the grey shaded area represent on these scatter plots?
11) Figure 1C: What does the blue shaded areas represent? Are the numbers 1-12 suppose to represent each month? The caption for Figure 1 in general is very confusing and jumps between Figure A, B and C.
12) Line 207: What is the difference between these minimum temperatures and those in the previous sentence?
13) Line 208: Was this “coldest month” always the same month?
14) Figure 2: This should be placed after the following paragraph. What are the legend labels in Figure 2A? This is the first time evapotranspiration has been mentioned. I would include a quick comment in the Methods section on how it was calculated. How much confidence is there in these values and the increasing general trend?
15) Line 228: What about the variability in precipitation and its increasing trend?
16) Line 233: Is the same method as described in the Methods section? I would delete this sentence or at least make it more clear for Figure S3.
17) Line 235: What is the difference between “daily averages” and “average winter temperatures” here?
18) Figure 3: What is the “blue box” described in the caption? What do the shaded parts of the trend lines represent? In panels C and D, it looks like both the trend and variability, with the variability lines hardly noticeable.
19) Line 254: “MinTmin” was shown not to be significant in Table S1.
20) Line 255: Based on Table S1, you have some indices with significant changes, but no significant trends in any of the spring indices or seasonal runoff variables? Is this correct? The same question for Figure 4.
21) Line 262: Are these trends described here different than those listed in Table S1? Seems like you are using two different data sets to examine similar trends. Table S1 is 1992-2022 and this analysis uses 1982-2022.
22) Line 278: Is “MaxTmax” a winter variable or a summer variable?
23) Figure 6: Any discussion on Figure 6A? Are the numbers on the x-axis representing months?
24) Line 320: The SOI values show an increased contribution from winter precipitation to streamflow. Any reason as to why this does not show up in your analysis of the data sets?
25) Line 321: Were there more mid-winter melting events found in the data sets? Would the shoulder seasons (spring and autumn) be the most affected by the warming rather than winter?
26) Line 355: As mentioned previously in the discussion, there was an increased contribution from winter precipitation with the SOI analysis. Figure S4 showed a decrease in total winter precipitation and an increase in total summer precipitation. Any comment on the different results? How would this influence the probability of droughts in this boreal catchment?
27) Line 372: “Shrinking snowpacks” As mentioned in previous comment, one analysis shows a greater contribution and the other, a non-significant decrease. Did you find that the snowpacks were shrinking?
1) Table S1: What is the difference between Tmin, Tmax and MaxTmax and MinTmin?
2) Table S1: Which 27 climate change indices were developed by the World Meteorological Organization out of the 33 variables listed?
3) Table S1: Should the growing season length be under the summer category instead of winter?
4) Figure S1: Are the dotted lines the average daily temperature in blue? It is difficult to distinguish between the three data sets in this figure.
5) Figure S2: For panel C and D, what does the pink shade represent?
6) Figure S3: If this is average daily temperature for each year, where is the “isolation of the winter period” in each figure?
7) Figure S5: What does the blue shade in the background of each figure represent?
Citation: https://doi.org/10.5194/hess-2024-337-RC2 -
RC3: 'Comment on hess-2024-337', Anonymous Referee #3, 15 Feb 2025
General comments
The authors studied trends in hydroclimatic extremes using a 40-year time series from the boreal Krycklan catchment in Sweden, in the context of 130 years of climate data from a nearby site. They looked at how different extreme climate indices changed and how they affected seasonal low flows in winter and summer. The authors also identified the best climate indices for predicting both summer and winter minimum flows. The authors have carried out an interesting study that is certainly scientifically relevant. I see the novelty both in the focus on a boreal region, where not so many studies exist, and in the robust testing of different predictors of minimum flow. Therefore, I believe that the study has the potential to be published in HESS. However, I have some comments listed below that I would like to be addressed.
Major comments
I think the research gaps and novelty could be better described. Although the entire introduction is clear and comprehensive, I miss the "so what" message. What is new in the study and how does it go beyond current knowledge?
L157: Please describe the linear regression model in more detail. The linear regression models also have some requirements for the input data, such as normality and uncorrelated predictors. Were these conditions met? If not, were the data transformed? Did the authors also consider other methods for detecting trends, such as Mann-Kendall's correlation (which can be used for non-normal distributions thanks to its ranking) or Sen's slopes? How might the interpretation of the results change if different methods of trend analysis were used?
L270, section Model of changes in runoff: I think the text in this section should be better supported by the results. For example, the authors state that AFDD<0 is the best explanatory variable, but I cannot see it in any of the figures (Fig. 5 only shows its relationship with winter/summer runoff). Perhaps you could at least show the correlation matrix of all predictors (heat map or similar) to support your statements.
My general concern is that more could be done to investigate the potential effects of individual climate indices on runoff. For example, how does climate variability affect the observed relationship between climate and hydrological variables, in particular the role of warm/cold, wet/dry or snow-poor/snow-rich years? I believe this direction would provide new insights into catchment behaviour.
While I like the idea of supporting the results of the time series data analysis with a more detailed analysis using stable water isotopes (especially for the Krycklan catchment with long-term and detailed isotope data), I feel that this part is not well connected to the other results and seems quite separate from the rest of the text. In addition, there is only one short paragraph and one figure in the results section relating to this part. I would encourage the authors to expand this part by adding more visualisations and results, and try to better connect it to the results of other analyses.
Specific comments
Section 2: It would be good to provide a map of the Krycklan catchment and its position within Europe/Sweden.
L110: Is there a reason why the study only uses data up to 2004 for the Stensele climate station? Later, in some figures, the data are plotted together with data from the Krycklan catchment (by different colours). Have the two time series been homogenised somehow in order to plot them together as a single time series? The distance between Krycklan and Stensele is 150 km, so one would expect different conditions at the two stations.
L134: What is meant by “Temperature intensity”?
L144: Since Table S1 is very important for the methods and results interpretation, I would prefer to put it directly to the main text to avoid jumps between the two files. I think it would be beneficial for the readers.
L154: How exactly has baseflow been calculated? If baseflow is only represented by minimum streamflow, I would suggest not calling it baseflow (use just minimum streamflow or similar). If you agree, this terminology would need to be changed throughout the text and figures.
Fig. 1: Please explain the grey background colours in the B) (probably confidence or prediction intervals). Same for Fig. 2, 3, 5 and 6. Besides, B) and C) descriptions in Fig. 1 caption are probably switched.
Fig. 4: Please describe what is in individual panels in the figure caption
Fig. 5: There are a couple of years considerably outside the prediction range. Is there any explanation for what might be the reason for this?
Fig. 5, caption: I think the resulting equation of the linear model should be placed in the main text together with some further description (see also my major comment related to the linear regression model).
L298: I would be rather cautious about attributing the results to baseflow, as baseflow was not calculated/simulated in your study (unless I missed something). An increase in Qmin in winter doesn't necessarily mean that baseflow also increases (although I agree that it is likely). If there are more melting periods in winter (higher fast flow component), Qmin will of course also increase, but the effect on baseflow may not be so straightforward. I'm not saying it's not true (as I expect the close connection of streamflow and baseflow in the Krycklan catchment, although I don't know the study area), but I would ask the authors to address this issue in the discussion (e.g. by adding some studies investigating this effect).
L324: Again, please consider whether you mean baseflow or Qmin.
L354-368: This paragraph is perhaps rather the introduction.
In my opinion, the conclusions can be formulated more specifically and with clear take-home messages. Additionally, the sentence in L372 (“shrinking snowpacks that melt earlier”) is not a conclusion coming from this study (although the statement itself is true), so please consider reformulation.
Fig. S2, S5: Please explain the background colours (S2, S5) and the red line (S3) in the figure captions.
Technical corrections
L8 and 18: Should there be “subsequent” rather than “preceding” (seasons; streamflow)?
L111: Please specify the exact time period to be consistent with the previous text.
L154-155: Coefficient of determination (rather coefficient of regression).
L174: The abbreviation SOI should be defined here for the seasonal origin index.
Citation: https://doi.org/10.5194/hess-2024-337-RC3
Viewed
HTML | XML | Total | Supplement | BibTeX | EndNote | |
---|---|---|---|---|---|---|
268 | 52 | 9 | 329 | 30 | 7 | 8 |
- HTML: 268
- PDF: 52
- XML: 9
- Total: 329
- Supplement: 30
- BibTeX: 7
- EndNote: 8
Viewed (geographical distribution)
Country | # | Views | % |
---|
Total: | 0 |
HTML: | 0 |
PDF: | 0 |
XML: | 0 |
- 1