the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 09 Aug 2022
Homogenizing Swiss snow depth series – Impact on decadal trends and extremes
Abstract. Our current knowledge on snow depth trends is based almost exclusively on these non-homogenized data.Long-term observations of deposited snow are well suited as indicator of climate change. However, like all other long-term observations, they are prone to inhomogeneities that can influence and change trends if not taken into account. We investigated the effects of removing inhomogeneities in the large network of Swiss snow depth observations on trends and extreme values of commonly used snow indices, such as snow days, seasonal averages or maximum snow depth in the period 1961–2021. For this task, three homogenization methods were applied: Climatol and HOMER, which use a median based adjustment method, and interpQM, which applies quantile based adjustments. All three were run using the same break points and input data. This allowed us to investigate and quantify the effects of these methods on the homogenization results. We found that all three methods agree well on trends in seasonal average snow depth, while differences are visible for seasonal maximum snow depth and the corresponding extreme values. Here, the quantile-based method performed slightly better than the two median-based methods, as it had the smallest number of stations outside the 95 % confidence interval for 50-year return periods of maximum snow depth. These differences are mainly caused by the way the reference series are selected. The combination of a high minimum correlation (>0.7) and restrictions in vertical (<300 m) and horizontal (<100 km) distances proved to be better suited than only using correlations or distances respectively as criteria. The adjustments removed all positive trends for snow days in the original data and strengthened the negative mean trend, especially for stations >1500 m. In addition, the number of significant negative stations was increased between 7–21 %, with the strongest changes at higher snow depths.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(1007 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1007 KB) - Metadata XML
- BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-715', Anonymous Referee #1, 13 Sep 2022
The manuscript submitted by Buchmann et al. is a nice contribution that complements the analysis performed by the same group of authors in the past years. I consider the manuscript of interest for the community, however I believe the methodology should be significantly better described. In particular, the fundamental step of the calculation of the adjustment factors and their comparison in terms of applied equations is missing. Moreover, the role played by autocorrelation in the analysis of the time series is not discussed. Additional comments are contained in the pdf file.
- AC1: 'Reply on RC1', Gernot Resch, 25 Oct 2022
-
RC2: 'Comment on egusphere-2022-715', Anonymous Referee #2, 20 Sep 2022
The paper by Buchmann et al. studies the impact of homogenization on snow series in Switzerland. It is a continuation of previous work on break detections. The application and testing of different homogenization approach for in-situ snow cover series is important for many climatological applications, but yet little researched. The authors have made a considerable contribution in this area previously and also with this paper.
The manuscript is well organized and, generally, well written. Figures focus strongly on displaying trend magnitudes as colours and not as x- or y-axes, which sometimes only allows a rough interpretation. Finally, there are two major concerns that would need be addressed before publication (see below). One is related to the methodology, which is not well described. Also it is difficult to understand how the authors arrived at some of the conclusion (e.g., in the abstract). The other concerns the non-random changes in trends.
# Major comments
- The description of the different homogenization approaches is very explanatory and vague. I find it hard to distinguish between what the methods do and need in general, and how the authors applied them. Maybe a table would help? I think it is important to understand if the approach works daily or monthly, what thresholds on distance (h and v) were used – btw, are they the same for all methods, it is unclear to me – what thresholds on correlations, etc. Are the adjustment factors applied additive or multiplicative? I think the authors conclusion in the abstract are based on their understanding, but for readers it is impossible to follow.
- Throughout the manuscript I was surprised to see that homogenization drove trends to be more negative. What are the reasons for this? In theory, I would expect non-climatic breaks to be random, especially since I think manual observing procedures have remained the same for snow depth for the past, or? So there should be no instrument bias. Inhomogeneities are then mostly related to observer changes or relocations, which should not be biased towards less snow consistently. But please correct me, if I’m wrong here. The authors also confirm this supposition in L194, where the percentages they give show more or less similar strengthening and weakening of trends. I think this issues deserves some more consideration. It might be partly related to how strength and direction of trends interact, especially when close to zero, where small changes in trends can lead to a change in sign. Could the authors provide a figure where they show explicitly how trends have change in a numerical fashion? Currently, all plots show trend magnitude with colours, where it is hard to judge values from. Also, since the authors have a larger dataset in the background (the homogeneous series they did not use here, but in previous works), they could compare inhomogenous vs homogeneous series in their fraction of positive/negative trends to check whether inhomogeneities were really biased to more snow.
# Minor comments
L16: A station cannot be significant, but its trend. Please adjust.
L19: Do you have a reference for that 50%? Does it refer to all of NH?
L41: Would be good to include Marcolini here, too.
L53: If I remember correctly, Marcolini did that.
L106: IQR is a widely used term in statistics for inter-quartile-range (note R not N in quantile/quartile), please use another abbreviation for clearness. And I think you mean that you split into different bins based on quantiles, and not split into inter quantile ranges, which are only numbers (differences between the quantiles).
L118: “Since it was decided…” : I do not understand what you mean, or who decided.
L127: The authors should explain why they choose exactly this specific approach to define “better” performance for R50HSmax, since this is not a trivial task.
L146: For what exactly did you use the KS-test?
L174(paragraph): I find the reasoning irritating in this paragraph. Are the trends in snow driven by absolute temperatures? Which the authors imply in L177. And then the statement in L179 is based on what source? Is it stating theory or factual evidence?
Figure 1: Could you please add an elevation distribution of the stations? And highlight in the distribution homogeneous and not.
Table1: Would be good to know how many stations are below 1500m and above 1500m. Also is the number constant across dHS5, 30 and 50? Also the numbers seem to not add up: Shouldn’t positive plus negative always be 100% for each column? It is for some columns, but not for all. And these should be consistent to the rows “pos to neg” and “neg to pos”, no?
Citation: https://doi.org/10.5194/egusphere-2022-715-RC2 - AC2: 'Reply on RC2', Gernot Resch, 09 Nov 2022
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-715', Anonymous Referee #1, 13 Sep 2022
The manuscript submitted by Buchmann et al. is a nice contribution that complements the analysis performed by the same group of authors in the past years. I consider the manuscript of interest for the community, however I believe the methodology should be significantly better described. In particular, the fundamental step of the calculation of the adjustment factors and their comparison in terms of applied equations is missing. Moreover, the role played by autocorrelation in the analysis of the time series is not discussed. Additional comments are contained in the pdf file.
- AC1: 'Reply on RC1', Gernot Resch, 25 Oct 2022
-
RC2: 'Comment on egusphere-2022-715', Anonymous Referee #2, 20 Sep 2022
The paper by Buchmann et al. studies the impact of homogenization on snow series in Switzerland. It is a continuation of previous work on break detections. The application and testing of different homogenization approach for in-situ snow cover series is important for many climatological applications, but yet little researched. The authors have made a considerable contribution in this area previously and also with this paper.
The manuscript is well organized and, generally, well written. Figures focus strongly on displaying trend magnitudes as colours and not as x- or y-axes, which sometimes only allows a rough interpretation. Finally, there are two major concerns that would need be addressed before publication (see below). One is related to the methodology, which is not well described. Also it is difficult to understand how the authors arrived at some of the conclusion (e.g., in the abstract). The other concerns the non-random changes in trends.
# Major comments
- The description of the different homogenization approaches is very explanatory and vague. I find it hard to distinguish between what the methods do and need in general, and how the authors applied them. Maybe a table would help? I think it is important to understand if the approach works daily or monthly, what thresholds on distance (h and v) were used – btw, are they the same for all methods, it is unclear to me – what thresholds on correlations, etc. Are the adjustment factors applied additive or multiplicative? I think the authors conclusion in the abstract are based on their understanding, but for readers it is impossible to follow.
- Throughout the manuscript I was surprised to see that homogenization drove trends to be more negative. What are the reasons for this? In theory, I would expect non-climatic breaks to be random, especially since I think manual observing procedures have remained the same for snow depth for the past, or? So there should be no instrument bias. Inhomogeneities are then mostly related to observer changes or relocations, which should not be biased towards less snow consistently. But please correct me, if I’m wrong here. The authors also confirm this supposition in L194, where the percentages they give show more or less similar strengthening and weakening of trends. I think this issues deserves some more consideration. It might be partly related to how strength and direction of trends interact, especially when close to zero, where small changes in trends can lead to a change in sign. Could the authors provide a figure where they show explicitly how trends have change in a numerical fashion? Currently, all plots show trend magnitude with colours, where it is hard to judge values from. Also, since the authors have a larger dataset in the background (the homogeneous series they did not use here, but in previous works), they could compare inhomogenous vs homogeneous series in their fraction of positive/negative trends to check whether inhomogeneities were really biased to more snow.
# Minor comments
L16: A station cannot be significant, but its trend. Please adjust.
L19: Do you have a reference for that 50%? Does it refer to all of NH?
L41: Would be good to include Marcolini here, too.
L53: If I remember correctly, Marcolini did that.
L106: IQR is a widely used term in statistics for inter-quartile-range (note R not N in quantile/quartile), please use another abbreviation for clearness. And I think you mean that you split into different bins based on quantiles, and not split into inter quantile ranges, which are only numbers (differences between the quantiles).
L118: “Since it was decided…” : I do not understand what you mean, or who decided.
L127: The authors should explain why they choose exactly this specific approach to define “better” performance for R50HSmax, since this is not a trivial task.
L146: For what exactly did you use the KS-test?
L174(paragraph): I find the reasoning irritating in this paragraph. Are the trends in snow driven by absolute temperatures? Which the authors imply in L177. And then the statement in L179 is based on what source? Is it stating theory or factual evidence?
Figure 1: Could you please add an elevation distribution of the stations? And highlight in the distribution homogeneous and not.
Table1: Would be good to know how many stations are below 1500m and above 1500m. Also is the number constant across dHS5, 30 and 50? Also the numbers seem to not add up: Shouldn’t positive plus negative always be 100% for each column? It is for some columns, but not for all. And these should be consistent to the rows “pos to neg” and “neg to pos”, no?
Citation: https://doi.org/10.5194/egusphere-2022-715-RC2 - AC2: 'Reply on RC2', Gernot Resch, 09 Nov 2022
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 307 | 125 | 16 | 448 | 2 | 2 |
- HTML: 307
- PDF: 125
- XML: 16
- Total: 448
- BibTeX: 2
- EndNote: 2
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Gernot Resch
Michael Begert
Stefan Brönnimann
Barbara Chimani
Wolfgang Schöner
Christoph Marty
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1007 KB) - Metadata XML