the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 17 Jun 2026
A 15 Year Climatology and Trend Study of Tropospheric Relative Humidity and Temperature over Switzerland based on Raman Lidar Measurements
Abstract. Water vapor is the most important greenhouse gas, yet large uncertainties remain regarding its evolution in a warming climate, particularly for relative humidity (RH). In this study we use 15 years (2010–2024) of nighttime Raman lidar measurements from the Raman Lidar for Meteorological Observations (RALMO) in Payerne, Switzerland, to investigate free-tropospheric RH changes under clear-sky conditions. We reprocessed and homogenized the data set and derived a climatology and seasonal trends between 3 and 10 km. The temperature climatology captures the expected annual cycle with an amplitude decreasing from about 10 K near the surface to 2 K near the tropopause. The RH climatology exhibits an annual cycle in the lower and upper troposphere and a semi-annual cycle in the mid-troposphere, reflecting the influence of large-scale circulation and moisture transport. Consistent tropospheric warming of approximately 1–3 K per decade is observed, with the strongest and statistically significant trends occurring in winter and summer. In contrast, RH trends are generally small and not statistically significant, suggesting that RH has remained approximately constant over Switzerland despite atmospheric warming. Our results imply that moisture supply is sufficient in the Alpine region to compensate for the atmosphere’s increasing capacity to hold water. This work is one of few long-term studies of RH in the free troposphere where such analyses remain limited by the scarcity of suitable datasets.
- Preprint
(3373 KB) - Metadata XML
- BibTeX
- EndNote
Status: closed
-
RC1: 'Comment on egusphere-2026-2823', Anonymous Referee #1, 26 Aug 2026
-
AC1: 'Reply on RC1', Alexander Haefele, 10 Oct 2026
We would like to thank the reviewer for carefully reading through our manuscript and for the positive report. We agree that comparisons with model data sets are a very interesting and pertinent continuation of this work, thank you for this suggestion.
Citation: https://doi.org/10.5194/egusphere-2026-2823-AC1
-
AC1: 'Reply on RC1', Alexander Haefele, 10 Oct 2026
-
RC2: 'Comment on egusphere-2026-2823', Anonymous Referee #2, 07 Sep 2026
This paper analyzes a 15 year temperature and water vapor dataset observed at Payerne Switzerland. The observations were primarily made by a Raman lidar, but the same analysis was performed on a full radiosonde datasets and a subset of the radiosondes that were selected to match the Raman lidar observations. They focused primarily on the trends observed between 3 and 10 km (i.e., 700 mb to 250 mb), separating the results by season. The authors wrote a well-organized and clear manuscript, and it was a pleasure to read. I only have a few questions / suggestions I would like to make:
- A key point to this paper is the calibration method applied to the Raman lidar temperature and RH fields. They point to a “ground-based internal calibration method developed by Jayaweera et al. (2025)”. Given the importance of this calibration to the paper, I would like to have it summarized in a new paragraph within this paper
- The key baseline variables measured by a Raman lidar are ambient temperature and water vapor mixing ratio. Thus, the RH field includes the uncertainties of both baseline variables. I think this paper would be much stronger if a similar trend analysis (e.g., Fig 12) could be performed with the water vapor mixing ratio. That should help make the point that even though temperature is increasing, the reason RH is not increasing is due to the decrease in the water vapor mixing ratio.
- Line 189 on filling the COVID gap: was the linear relationship applied to temperature and RH simultaneously? Or was the temperature gap filled first, and then the RH gap filled? Or was the temperature and water vapor mixing ratio gaps filled, and then the RH was computed? (This connects with the previous point about mixing ratio being a more baseline observation than RH)
- Paragraph starting line 235: November is also one of the more poorly sampled months; is it possible that the increase in the RH seen in November is due to sampling?
- Line 249: The authors point out that there is more temperature variability from the RALMO obs due to instrument noise. Instrument noise would also affect the water vapor mixing ratio, and thus RH would have two instrumental sources of noise. The authors have nicely indicated the amount of temperature variability that comes from the lidar here, but how much RH variability comes from the lidar noise?
- Fig 8: Do you have a hypothesis why the RALMO observations show a distinct lower amount of variability throughout the entire troposphere in November? Does the matched radiosonde dataset show this reduction also?
- Line 413: I do not agree with the statement that the clear sky sampling bias is negligible (my interpretation of “do not depend critically”). In Fig 11 (especially) and also Fig 12, the all 95% confidence interval is markedly wider for the “all sonde” than for the lidar. Even the “sonde selected” results generally have wider 95% confidence intervals than the lidar, and often much larger. The authors need to be more careful with this particular conclusion, as what they have written is not supported by the analysis.
Minor points:
- Line 52 “The Raman Lidar for …” should be the start of a new paragraph
- Fig 4: please explain those few lidar points that have RH significantly higher than 100%. I have presumed throughout this document that RH is reported as RH with respect to liquid at all heights; is this correct? (It should be stated in the paper regardless).
- Line 290: Fall does not need to be capitalized
Citation: https://doi.org/10.5194/egusphere-2026-2823-RC2 -
AC2: 'Reply on RC2', Alexander Haefele, 10 Oct 2026
We thank the reviewer for the careful reading of our manuscript and for the constructive comments. Each point has been thoughtfully addressed, and we believe these revisions have enhanced the manuscript. Thank you for your contribution to this work.
Please note that all line numbers in our responses below refer to the revised manuscript with track changes submitted separately.
A key point to this paper is the calibration method applied to the Raman lidar temperature and RH fields. They point to a “ground-based internal calibration method developed by Jayaweera et al. (2025)”. Given the importance of this calibration to the paper, I would like to have it summarized in a new paragraph within this paper.
We agree that the calibration method deserves a fuller description given its importance to the analysis. Therefore, a paragraph summarizing the solar background calibration method has been added. Please refer to lines 121–132.
The key baseline variables measured by a Raman lidar are ambient temperature and water vapor mixing ratio. Thus, the RH field includes the uncertainties of both baseline variables. I think this paper would be much stronger if a similar trend analysis (e.g., Fig. 12) could be performed with the water vapor mixing ratio. That should help make the point that even though temperature is increasing, the reason RH is not increasing is due to the decrease in the water vapor mixing ratio.
We agree that water vapour mixing ratio is a fundamental measure of atmospheric moisture and that its temporal evolution provides useful context for interpreting the RH results. However, the focus of the present study is the climatology and trends of RH. Our OEM algorithm retrieves temperature and RH directly and simultaneously from the raw lidar signals, rather than first retrieving water vapour mixing ratio and subsequently deriving RH. Consequently, a mixing ratio trend analysis is not a direct by-product of the retrievals presented here.
We also note that a height-resolved analysis of water vapour trends from RALMO has already been presented by Hicks-Jalali et al. (2020), based on 11.5 years of measurements. Their results showed statistically significant increases in precipitable water vapour and specific humidity over portions of the troposphere, consistent with Clausius-Clapeyron scaling under a warming climate. We refer to this work in the manuscript at lines 54 and 401ff.
As the RALMO record continues to lengthen, extending this analysis using the full time series would provide a natural avenue for future work.
Line 189 on filling the COVID gap: was the linear relationship applied to temperature and RH simultaneously? Or was the temperature gap filled first, and then the RH gap filled? Or was the temperature and water vapor mixing ratio gaps filled, and then the RH was computed? (This connects with the previous point about mixing ratio being a more baseline observation than RH.)
The gap was filled independently for temperature and RH. As noted in our response to the previous comment, our OEM algorithm retrieves RH directly from the raw measurements, so mixing ratio is not readily available from the reprocessed data set. Given that the trends are only to a very small degree sensitive to this single missing year, we consider the independent treatment of the two fields to be adequate.
We amended the text very slightly; please refer to line 197.
Paragraph starting line 235: November is also one of the more poorly sampled months; is it possible that the increase in the RH seen in November is due to sampling?
We do not believe the November feature to be a sampling artifact. Although November is among the more sparsely sampled months in the RALMO record, the elevated RH values observed in November are also present in the climatology computed from the full radiosonde record, which is sampled uniformly throughout the year and is therefore not subject to the same reduction in November coverage.
The agreement between these two independently sampled datasets therefore suggests that the November increase represents a genuine seasonal feature rather than an artifact arising from the lidar sampling. We have added a sentence at line 252 to clarify this.
Line 249: The authors point out that there is more temperature variability from the RALMO observations due to instrument noise. Instrument noise would also affect the water vapor mixing ratio, and thus RH would have two instrumental sources of noise. The authors have nicely indicated the amount of temperature variability that comes from the lidar here, but how much RH variability comes from the lidar noise?
Regarding the contribution of lidar noise to the observed RH variability, this is already quantified at line 273. There, we show that, relative to the radiosonde dataset with identical sampling, the RALMO measurements exhibit only about 10% greater RH variability in the upper summer troposphere. This indicates that the excess RH variability attributable to lidar noise is small over most of the troposphere.
We made a small amendment of the text on line 274.
Fig. 8: Do you have a hypothesis why the RALMO observations show a distinct lower amount of variability throughout the entire troposphere in November? Does the matched radiosonde dataset show this reduction also?
The reduced variability in November is largely a consequence of how the variability is expressed. As noted at line 256, RH variability is reported in relative units, normalized by the mean RH. Because the climatology shows higher RH in November than in the adjacent months, a comparable absolute spread corresponds to a smaller relative variability.
The same reduction was present in the matched radiosonde dataset, confirming that this was not an artifact of the lidar measurements.
Line 413: I do not agree with the statement that the clear-sky sampling bias is negligible (my interpretation of “do not depend critically”). In Fig. 11 (especially) and also Fig. 12, the 95% confidence interval is markedly wider for the “all sonde” than for the lidar. Even the “sonde selected” results generally have wider 95% confidence intervals than the lidar, and often much larger. The authors need to be more careful with this particular conclusion, as what they have written is not supported by the analysis.
It is possible that our presentation of Figs. 11 and 12 was not sufficiently clear, as the confidence intervals shown for the lidar are in fact the wider of the two, rather than those of the all-sonde or sonde-selected datasets.
If we have understood the reviewer’s point in the intended sense, then our reasoning is as follows: the lidar confidence intervals are broad because of the instrumental noise in the RALMO measurements, and it is relative to these comparatively large intervals that the clear-sky sampling bias does not become a limiting factor.
We agree with the reviewer that this is a statement about the current uncertainty level rather than a general one, and that as the confidence intervals narrow with a longer record, the clear-sky sampling bias would become increasingly important. We have rephrased the passage at lines 428 to 433 to make this conditional nature explicit.
Minor comments
Line 52: “The Raman Lidar for ...” should be the start of a new paragraph.
Fixed.
Fig. 4: Please explain those few lidar points that have RH significantly higher than 100%. I have presumed throughout this document that RH is reported as RH with respect to liquid at all heights; is this correct? (It should be stated in the paper regardless.)
The few points exceeding 100% result from a combination of measurement noise in the retrieved profiles and, to some extent, genuine supersaturation with respect to liquid water. We have added a brief explanation of this to the discussion of Fig. 4 at line 203.
The reviewer’s presumption is correct: RH is reported with respect to liquid water at all altitudes, and we have now stated this explicitly at line 96.
Line 290: Fall does not need to be capitalized.
Fixed.
Citation: https://doi.org/10.5194/egusphere-2026-2823-AC2
Status: closed
-
RC1: 'Comment on egusphere-2026-2823', Anonymous Referee #1, 26 Aug 2026
This manuscript presents a carefully conducted study that explores mid and upper tropospheric trends in temperature and relative humidity. I found the methodology utilizing Raman lidar and radiosondes to be appropriate, the analysis to be thorough, and the conclusions to not only be well supported by the results but support many papers exploring similar trends. The manuscript is also clearly written and logically structured. I did not identify any major or minor scientific concerns that would require revision and I find the manuscript to be a strong contribution to the literature and have no substantive comments for the authors. One potential direction for future research that arises from this study is determining how reanalysis or model outputs compare to trends identified here. I do not consider this necessary for the present manuscript, but I think exploring that in a future study could provide more dates to help fill in observational gaps in the analysis.
Citation: https://doi.org/10.5194/egusphere-2026-2823-RC1 -
AC1: 'Reply on RC1', Alexander Haefele, 10 Oct 2026
We would like to thank the reviewer for carefully reading through our manuscript and for the positive report. We agree that comparisons with model data sets are a very interesting and pertinent continuation of this work, thank you for this suggestion.
Citation: https://doi.org/10.5194/egusphere-2026-2823-AC1
-
AC1: 'Reply on RC1', Alexander Haefele, 10 Oct 2026
-
RC2: 'Comment on egusphere-2026-2823', Anonymous Referee #2, 07 Sep 2026
This paper analyzes a 15 year temperature and water vapor dataset observed at Payerne Switzerland. The observations were primarily made by a Raman lidar, but the same analysis was performed on a full radiosonde datasets and a subset of the radiosondes that were selected to match the Raman lidar observations. They focused primarily on the trends observed between 3 and 10 km (i.e., 700 mb to 250 mb), separating the results by season. The authors wrote a well-organized and clear manuscript, and it was a pleasure to read. I only have a few questions / suggestions I would like to make:
- A key point to this paper is the calibration method applied to the Raman lidar temperature and RH fields. They point to a “ground-based internal calibration method developed by Jayaweera et al. (2025)”. Given the importance of this calibration to the paper, I would like to have it summarized in a new paragraph within this paper
- The key baseline variables measured by a Raman lidar are ambient temperature and water vapor mixing ratio. Thus, the RH field includes the uncertainties of both baseline variables. I think this paper would be much stronger if a similar trend analysis (e.g., Fig 12) could be performed with the water vapor mixing ratio. That should help make the point that even though temperature is increasing, the reason RH is not increasing is due to the decrease in the water vapor mixing ratio.
- Line 189 on filling the COVID gap: was the linear relationship applied to temperature and RH simultaneously? Or was the temperature gap filled first, and then the RH gap filled? Or was the temperature and water vapor mixing ratio gaps filled, and then the RH was computed? (This connects with the previous point about mixing ratio being a more baseline observation than RH)
- Paragraph starting line 235: November is also one of the more poorly sampled months; is it possible that the increase in the RH seen in November is due to sampling?
- Line 249: The authors point out that there is more temperature variability from the RALMO obs due to instrument noise. Instrument noise would also affect the water vapor mixing ratio, and thus RH would have two instrumental sources of noise. The authors have nicely indicated the amount of temperature variability that comes from the lidar here, but how much RH variability comes from the lidar noise?
- Fig 8: Do you have a hypothesis why the RALMO observations show a distinct lower amount of variability throughout the entire troposphere in November? Does the matched radiosonde dataset show this reduction also?
- Line 413: I do not agree with the statement that the clear sky sampling bias is negligible (my interpretation of “do not depend critically”). In Fig 11 (especially) and also Fig 12, the all 95% confidence interval is markedly wider for the “all sonde” than for the lidar. Even the “sonde selected” results generally have wider 95% confidence intervals than the lidar, and often much larger. The authors need to be more careful with this particular conclusion, as what they have written is not supported by the analysis.
Minor points:
- Line 52 “The Raman Lidar for …” should be the start of a new paragraph
- Fig 4: please explain those few lidar points that have RH significantly higher than 100%. I have presumed throughout this document that RH is reported as RH with respect to liquid at all heights; is this correct? (It should be stated in the paper regardless).
- Line 290: Fall does not need to be capitalized
Citation: https://doi.org/10.5194/egusphere-2026-2823-RC2 -
AC2: 'Reply on RC2', Alexander Haefele, 10 Oct 2026
We thank the reviewer for the careful reading of our manuscript and for the constructive comments. Each point has been thoughtfully addressed, and we believe these revisions have enhanced the manuscript. Thank you for your contribution to this work.
Please note that all line numbers in our responses below refer to the revised manuscript with track changes submitted separately.
A key point to this paper is the calibration method applied to the Raman lidar temperature and RH fields. They point to a “ground-based internal calibration method developed by Jayaweera et al. (2025)”. Given the importance of this calibration to the paper, I would like to have it summarized in a new paragraph within this paper.
We agree that the calibration method deserves a fuller description given its importance to the analysis. Therefore, a paragraph summarizing the solar background calibration method has been added. Please refer to lines 121–132.
The key baseline variables measured by a Raman lidar are ambient temperature and water vapor mixing ratio. Thus, the RH field includes the uncertainties of both baseline variables. I think this paper would be much stronger if a similar trend analysis (e.g., Fig. 12) could be performed with the water vapor mixing ratio. That should help make the point that even though temperature is increasing, the reason RH is not increasing is due to the decrease in the water vapor mixing ratio.
We agree that water vapour mixing ratio is a fundamental measure of atmospheric moisture and that its temporal evolution provides useful context for interpreting the RH results. However, the focus of the present study is the climatology and trends of RH. Our OEM algorithm retrieves temperature and RH directly and simultaneously from the raw lidar signals, rather than first retrieving water vapour mixing ratio and subsequently deriving RH. Consequently, a mixing ratio trend analysis is not a direct by-product of the retrievals presented here.
We also note that a height-resolved analysis of water vapour trends from RALMO has already been presented by Hicks-Jalali et al. (2020), based on 11.5 years of measurements. Their results showed statistically significant increases in precipitable water vapour and specific humidity over portions of the troposphere, consistent with Clausius-Clapeyron scaling under a warming climate. We refer to this work in the manuscript at lines 54 and 401ff.
As the RALMO record continues to lengthen, extending this analysis using the full time series would provide a natural avenue for future work.
Line 189 on filling the COVID gap: was the linear relationship applied to temperature and RH simultaneously? Or was the temperature gap filled first, and then the RH gap filled? Or was the temperature and water vapor mixing ratio gaps filled, and then the RH was computed? (This connects with the previous point about mixing ratio being a more baseline observation than RH.)
The gap was filled independently for temperature and RH. As noted in our response to the previous comment, our OEM algorithm retrieves RH directly from the raw measurements, so mixing ratio is not readily available from the reprocessed data set. Given that the trends are only to a very small degree sensitive to this single missing year, we consider the independent treatment of the two fields to be adequate.
We amended the text very slightly; please refer to line 197.
Paragraph starting line 235: November is also one of the more poorly sampled months; is it possible that the increase in the RH seen in November is due to sampling?
We do not believe the November feature to be a sampling artifact. Although November is among the more sparsely sampled months in the RALMO record, the elevated RH values observed in November are also present in the climatology computed from the full radiosonde record, which is sampled uniformly throughout the year and is therefore not subject to the same reduction in November coverage.
The agreement between these two independently sampled datasets therefore suggests that the November increase represents a genuine seasonal feature rather than an artifact arising from the lidar sampling. We have added a sentence at line 252 to clarify this.
Line 249: The authors point out that there is more temperature variability from the RALMO observations due to instrument noise. Instrument noise would also affect the water vapor mixing ratio, and thus RH would have two instrumental sources of noise. The authors have nicely indicated the amount of temperature variability that comes from the lidar here, but how much RH variability comes from the lidar noise?
Regarding the contribution of lidar noise to the observed RH variability, this is already quantified at line 273. There, we show that, relative to the radiosonde dataset with identical sampling, the RALMO measurements exhibit only about 10% greater RH variability in the upper summer troposphere. This indicates that the excess RH variability attributable to lidar noise is small over most of the troposphere.
We made a small amendment of the text on line 274.
Fig. 8: Do you have a hypothesis why the RALMO observations show a distinct lower amount of variability throughout the entire troposphere in November? Does the matched radiosonde dataset show this reduction also?
The reduced variability in November is largely a consequence of how the variability is expressed. As noted at line 256, RH variability is reported in relative units, normalized by the mean RH. Because the climatology shows higher RH in November than in the adjacent months, a comparable absolute spread corresponds to a smaller relative variability.
The same reduction was present in the matched radiosonde dataset, confirming that this was not an artifact of the lidar measurements.
Line 413: I do not agree with the statement that the clear-sky sampling bias is negligible (my interpretation of “do not depend critically”). In Fig. 11 (especially) and also Fig. 12, the 95% confidence interval is markedly wider for the “all sonde” than for the lidar. Even the “sonde selected” results generally have wider 95% confidence intervals than the lidar, and often much larger. The authors need to be more careful with this particular conclusion, as what they have written is not supported by the analysis.
It is possible that our presentation of Figs. 11 and 12 was not sufficiently clear, as the confidence intervals shown for the lidar are in fact the wider of the two, rather than those of the all-sonde or sonde-selected datasets.
If we have understood the reviewer’s point in the intended sense, then our reasoning is as follows: the lidar confidence intervals are broad because of the instrumental noise in the RALMO measurements, and it is relative to these comparatively large intervals that the clear-sky sampling bias does not become a limiting factor.
We agree with the reviewer that this is a statement about the current uncertainty level rather than a general one, and that as the confidence intervals narrow with a longer record, the clear-sky sampling bias would become increasingly important. We have rephrased the passage at lines 428 to 433 to make this conditional nature explicit.
Minor comments
Line 52: “The Raman Lidar for ...” should be the start of a new paragraph.
Fixed.
Fig. 4: Please explain those few lidar points that have RH significantly higher than 100%. I have presumed throughout this document that RH is reported as RH with respect to liquid at all heights; is this correct? (It should be stated in the paper regardless.)
The few points exceeding 100% result from a combination of measurement noise in the retrieved profiles and, to some extent, genuine supersaturation with respect to liquid water. We have added a brief explanation of this to the discussion of Fig. 4 at line 203.
The reviewer’s presumption is correct: RH is reported with respect to liquid water at all altitudes, and we have now stated this explicitly at line 96.
Line 290: Fall does not need to be capitalized.
Fixed.
Citation: https://doi.org/10.5194/egusphere-2026-2823-AC2
Data sets
Tropospheric Temperature and Relative Humidity Profile Record from Raman Lidar over Payerne (Switzerland) Y. V. Jayaweera et al. https://doi.org/10.5281/zenodo.20034242
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 171 | 142 | 48 | 361 | 44 | 41 |
- HTML: 171
- PDF: 142
- XML: 48
- Total: 361
- BibTeX: 44
- EndNote: 41
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This manuscript presents a carefully conducted study that explores mid and upper tropospheric trends in temperature and relative humidity. I found the methodology utilizing Raman lidar and radiosondes to be appropriate, the analysis to be thorough, and the conclusions to not only be well supported by the results but support many papers exploring similar trends. The manuscript is also clearly written and logically structured. I did not identify any major or minor scientific concerns that would require revision and I find the manuscript to be a strong contribution to the literature and have no substantive comments for the authors. One potential direction for future research that arises from this study is determining how reanalysis or model outputs compare to trends identified here. I do not consider this necessary for the present manuscript, but I think exploring that in a future study could provide more dates to help fill in observational gaps in the analysis.