the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Sep 2025
Evolution of nonstationary hydrological drought characteristics in the UK under warming
Abstract. Although the United Kingdom (UK) is relatively wet, there is an increasing awareness of the impacts of droughts, and an expectation that droughts will become worse in the future. This has motivated studies that have developed projections of future UK drought characteristics. To date, however, very few have addressed future changes in terms of probability of occurrence, and none have quantified the evolution of rare nonstationary hydrological drought characteristics under different warming conditions. This study investigates future changes in the hydrological drought characteristics under varying global warming levels (1.5 °C, 2 °C, and 3 °C), using nonstationary extreme value analysis combined with a Bayesian uncertainty framework across 200 river catchments in the UK. The analysis utilizes the enhanced future Flows and Groundwater (eFLaG) dataset, which is based on the most recent UKCP18 climate projections, and incorporates outputs from four hydrological models (G2G, PDM, GR4J, and GR6J). The findings indicate that rising temperatures will significantly influence future drought duration, severity, and intensity across a majority of catchments, with rare droughts (return period of 100–500 years) projected to be more severe in all seasons, particularly in the southern UK. Further, relatively frequent summer droughts (return periods of 10 years) are expected to become shorter but more severe and intense, particularly at higher warming. We observe notable differences between stationary and nonstationary return periods across seasons, with the change becoming more pronounced at longer return periods, particularly for drought severity. Although the trends remain consistent across models under stationary and nonstationary conditions, the results underscore the role of rarity, nonstationarity, and seasonal controls on the future evolution of hydrological droughts in the region.
-
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
(1674 KB)
-
Supplement
(7802 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1674 KB) - Metadata XML
-
Supplement
(7802 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4096', Anonymous Referee #1, 18 Sep 2025
- AC1: 'Reply on RC1', Srinidhi Jha, 10 Dec 2025
-
RC2: 'Comment on egusphere-2025-4096', Anonymous Referee #2, 01 Oct 2025
- The paper is well-written and clearly mentions the research gap. It would be more interesting if the authors kept the objective as a point order (consistent) before concluding the introduction, so that the reader remains clear about the objective before moving on to the methodology.
- The author highlighted the role of seasonal controls in future drought. In lines 45 and the abstract, it has been highlighted, but it is unclear whether it is also considered an objective of the study. However, it's clear from the results section onwards, and this is indeed very interesting. However, the authors have not highlighted the importance of seasonality in relation to drought in the research gap and have not cited any relevant literature.
- Can the author justify why they have selected the 90 th percentile threshold? A suggestion to check with an 80th percentile threshold.
- The majority of studies have considered the drought identification (threshold) on the total observation period, so using a baseline period here is really questionable. Please justify. For future data, the threshold should also vary, so the entire period should be considered instead of just the baseline.
- Line 146 Repetition about the number of catchments may be deleted, as it has been clearly stated beforehand.
- Line 186: Please check if the null hypothesis is rejected for a P-value exceeding 0.05; is this the case?
Please clarify the exact null hypothesis in accordance with the study and cite the relevant literature.
Present a table which is significant, which is not inthe supplementary - Line 198, there is an error in the terminology used in the expression and in the description.
- Line 232: What is the moving window size with respect to ± how many days?
- Line 240: Cite literature
- Line 382-385: Also, the intensity is a function of both duration and severity, so that might be the reason behind the smaller variability?
- Figure 2: Mention unit for duration, intensity and severity
Citation: https://doi.org/10.5194/egusphere-2025-4096-RC2 - AC2: 'Reply on RC2', Srinidhi Jha, 10 Dec 2025
-
RC3: 'Comment on egusphere-2025-4096', Anonymous Referee #3, 29 Oct 2025
Review of "Evolution of nonstationary hydrological drought characteristics in the UK under warming" by Jha et al.
This manuscript proposes an analysis of the probability of occurence of rare hydrological droughts in the UK under different warming levels. It makes use of state-of-the-art multimodel hydrological projections over the 21st century and of an extreme value analysis of drought characteristics (duration, intensity, severity). Note that hydrological drought are defined through daily anomalies with respect to an average daily regime and the manuscript topic is therefore not about low-flows. The manuscript is well written and well organised. Methods, results, and corresponding conclusions are sound. Moreover, such an analysis is timely when climate change adaptation is more and more organised around Global or Regional Warming Levels (see e.g. Sauquet et al., 2025). I have however one concern about the seasonal approach chosen to present results. It is detailed -- with suggestions of improvements -- along another general comment below. Specific comments are also detailed afterwards. I would therefore recommend a revision to be made on this point before publication.
General comments
- L244-249: This is where the methodological choices seem rather strange. I do not understand why results are artificially broken by seasons after an event-based analysis. Indeed, any physical continuity across seasons is lost while this continuity is of uttermost importance when analysing seasonal shifts in drought development. Furthermore, analysing seasons separately leads invitably to an artifical upper bound for duration (and therefore severity). I would strongly suggest considering instead e.g. the proportion of each individual event in each season (possibly considering also different consecutive years/seasons for a single multiyear event). This would only slightly alter the analysis and bring much more insight into drought changes.
- The warming level analysis is rather attractive, but associated uncertainties are barely discussed in the manuscript although they are crucial for adaptation purposes. Some specfic comments below detail a few aspects of that, but the main question arising here is the relative importance of the diversity of climate projections, the diversity of hydrological models, and GEV parameter estimation. I may have missed corresponding explanations, but it is unclear how pooling is done across climate projections (and or not across hydrological models) to derive GEV estimates. Making hypotheses even clearer is a major point of potential improvement for the manuscript.
Specific comments
- L146-150: Does eFlag use CHESS-SCAPE as forcings for hyrological models? Please make it clear.
- L159-161: How is the daily temperature anomaly defined in CHESS-SCAPE? What is the reference period? How is seasonality taken into account? What is the spatial scale considered for computing the anomalies: local, UK, global? Please detail the answers explicitely in the manuscript.
- L 220-224: Please define exactly what these warming levels refer to. Are they Global Warming Levels (GWLs)? Are they Regional (UK) Warming Levels (RWLs)? In the first case, how GWLs translate into RWLs?
- L238-251: How does this pooling procedure compare with the Sequent Peak Algorithm (SPA) traditionally used with a fixed threshold?
- L242: What does "standard" refer to? 30 days sound already quite long for a drought event, even with a daily varying threshold. Please comment on that. Plus, this introduces a hard lower bound for duration, which may hide some signal on "flash droughts". I would therefore recommend not censuring a priori such short events which might (depending on catchment storage) bring in some extreme values of e.g. intensity.
- Section 3.1. and Fig. 2.: Fig. 2 presents strong fluctuations of stationarity properties across warming levels with e.g. the number of nonstationary catchments declining from 1.5°C to 2°C and increasing to 3°C. This demonstrates the limits of pattern scaling properties and what is striking is that this is barely commented in the corresponding text of Sect. 3.1. Please at least add such comments on that, as pattern scaling is indeed one of the foundations of this manuscript. What is also missing here is the corresponding spatial patterns of such nonstationarity, which would be interesting for understanding the limits of pattern scaling across the UK.
- L.288: Fig. S1 is rather required in the main text in my view.
- Fig. 3: This is clearly unreadable as such, for several reasons. One, the color scale is non linear, which is definitely against perceptual rules (Hawkins, 2015 ; Stoelze and Stein, 2021). Two, maps are way too small. Three, using catchment surface as the support for colors is not appropriate for such small figure dimensions, as it perceptually highlights only large catchments. I would therefore strongly suggest using shapes at the outlet of each catchment. Four, colorscale label sizes are not homogeneous across facets, and the colorscale titles "location paramater" are redundant.
- Fig. 4: This is again much too small for being readable. And again, I would strongly recommend using e.g. disc shapes, and also using grey instead of black for UK costlines. A discrete color scale might also be more effective.
- Fig. 5: Please confirm (in the legend) that boxplots show differences across catchments only (and not across climate/hydrological models).
- L371-372: What can the reader refer to when reading that rarer droughts are accompanied by large variability? Across what? Spatial? Parameter estimation? Other?
- L421: Please provide factural elements to comfort the assertion of "robust estimates of uncertainty".
References
Hawkins, E. (2015) Graphics: Scrap rainbow colour scales. Nature, 519, 291. https://doi.org/10.1038/519291d
Sauquet et al. (2025), Évolution de l’hydrologie de surface en France par niveau de réchauffement, https://doi.org/10.57745/MN29RG, Recherche Data Gouv, V5.
Stoelzle, M. & Stein, L. (2021) Rainbow color map distorts and misleads research in hydrology -- guidance for better visualizations and science communication. Hydrology and Earth System Sciences, 25, 4549-4565. https://doi.org/10.5194/hess-25-4549-2021
Citation: https://doi.org/10.5194/egusphere-2025-4096-RC3 - AC3: 'Reply on RC3', Srinidhi Jha, 10 Dec 2025
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2025-4096', Anonymous Referee #1, 18 Sep 2025
Title: Evolution of nonstationary hydrological drought characteristics in the UK under warming
Recommendation: Accept after corrections
Explain the non-stationarity in the hydrological drought time series. How the future groundwater estimates are calculated and how accurate it is?
Line 35: 1.2 deg is the how many years average?
Line 80-81: ….”transient changes in low-flows characteristics”; what is the meaning of the transient changes here?
Line 113: What is “transient” is not clear from the introduction?
Line 117-118: “we aim to capture the full spectrum of possible future hydrological drought conditions under different climatic conditions.”
Line 133 – 135: “It should be noted that all 12 ensemble members originate from the same model framework and are based on the high emissions scenario (RCP8.5).” How it is same model framework? Please rephrase or write proper explanation for these lines?
Line 148: “recently developed CHESS-SCAPE” what do you mean by “recently developed”
Figure 1: which approach is more suitable for the drought identification; variable threshold or stationary approach? Have authors included the justification and applicability for these methods?
Conclusions:
Authors are advised to write the conclusions with the focus on the comparative analysis of the drought occurrence in the baseline and future periods under different warmings. Which can help in the framing/modifying the policy for the future dryness events.
Citation: https://doi.org/10.5194/egusphere-2025-4096-RC1 - AC1: 'Reply on RC1', Srinidhi Jha, 10 Dec 2025
-
RC2: 'Comment on egusphere-2025-4096', Anonymous Referee #2, 01 Oct 2025
- The paper is well-written and clearly mentions the research gap. It would be more interesting if the authors kept the objective as a point order (consistent) before concluding the introduction, so that the reader remains clear about the objective before moving on to the methodology.
- The author highlighted the role of seasonal controls in future drought. In lines 45 and the abstract, it has been highlighted, but it is unclear whether it is also considered an objective of the study. However, it's clear from the results section onwards, and this is indeed very interesting. However, the authors have not highlighted the importance of seasonality in relation to drought in the research gap and have not cited any relevant literature.
- Can the author justify why they have selected the 90 th percentile threshold? A suggestion to check with an 80th percentile threshold.
- The majority of studies have considered the drought identification (threshold) on the total observation period, so using a baseline period here is really questionable. Please justify. For future data, the threshold should also vary, so the entire period should be considered instead of just the baseline.
- Line 146 Repetition about the number of catchments may be deleted, as it has been clearly stated beforehand.
- Line 186: Please check if the null hypothesis is rejected for a P-value exceeding 0.05; is this the case?
Please clarify the exact null hypothesis in accordance with the study and cite the relevant literature.
Present a table which is significant, which is not inthe supplementary - Line 198, there is an error in the terminology used in the expression and in the description.
- Line 232: What is the moving window size with respect to ± how many days?
- Line 240: Cite literature
- Line 382-385: Also, the intensity is a function of both duration and severity, so that might be the reason behind the smaller variability?
- Figure 2: Mention unit for duration, intensity and severity
Citation: https://doi.org/10.5194/egusphere-2025-4096-RC2 - AC2: 'Reply on RC2', Srinidhi Jha, 10 Dec 2025
-
RC3: 'Comment on egusphere-2025-4096', Anonymous Referee #3, 29 Oct 2025
Review of "Evolution of nonstationary hydrological drought characteristics in the UK under warming" by Jha et al.
This manuscript proposes an analysis of the probability of occurence of rare hydrological droughts in the UK under different warming levels. It makes use of state-of-the-art multimodel hydrological projections over the 21st century and of an extreme value analysis of drought characteristics (duration, intensity, severity). Note that hydrological drought are defined through daily anomalies with respect to an average daily regime and the manuscript topic is therefore not about low-flows. The manuscript is well written and well organised. Methods, results, and corresponding conclusions are sound. Moreover, such an analysis is timely when climate change adaptation is more and more organised around Global or Regional Warming Levels (see e.g. Sauquet et al., 2025). I have however one concern about the seasonal approach chosen to present results. It is detailed -- with suggestions of improvements -- along another general comment below. Specific comments are also detailed afterwards. I would therefore recommend a revision to be made on this point before publication.
General comments
- L244-249: This is where the methodological choices seem rather strange. I do not understand why results are artificially broken by seasons after an event-based analysis. Indeed, any physical continuity across seasons is lost while this continuity is of uttermost importance when analysing seasonal shifts in drought development. Furthermore, analysing seasons separately leads invitably to an artifical upper bound for duration (and therefore severity). I would strongly suggest considering instead e.g. the proportion of each individual event in each season (possibly considering also different consecutive years/seasons for a single multiyear event). This would only slightly alter the analysis and bring much more insight into drought changes.
- The warming level analysis is rather attractive, but associated uncertainties are barely discussed in the manuscript although they are crucial for adaptation purposes. Some specfic comments below detail a few aspects of that, but the main question arising here is the relative importance of the diversity of climate projections, the diversity of hydrological models, and GEV parameter estimation. I may have missed corresponding explanations, but it is unclear how pooling is done across climate projections (and or not across hydrological models) to derive GEV estimates. Making hypotheses even clearer is a major point of potential improvement for the manuscript.
Specific comments
- L146-150: Does eFlag use CHESS-SCAPE as forcings for hyrological models? Please make it clear.
- L159-161: How is the daily temperature anomaly defined in CHESS-SCAPE? What is the reference period? How is seasonality taken into account? What is the spatial scale considered for computing the anomalies: local, UK, global? Please detail the answers explicitely in the manuscript.
- L 220-224: Please define exactly what these warming levels refer to. Are they Global Warming Levels (GWLs)? Are they Regional (UK) Warming Levels (RWLs)? In the first case, how GWLs translate into RWLs?
- L238-251: How does this pooling procedure compare with the Sequent Peak Algorithm (SPA) traditionally used with a fixed threshold?
- L242: What does "standard" refer to? 30 days sound already quite long for a drought event, even with a daily varying threshold. Please comment on that. Plus, this introduces a hard lower bound for duration, which may hide some signal on "flash droughts". I would therefore recommend not censuring a priori such short events which might (depending on catchment storage) bring in some extreme values of e.g. intensity.
- Section 3.1. and Fig. 2.: Fig. 2 presents strong fluctuations of stationarity properties across warming levels with e.g. the number of nonstationary catchments declining from 1.5°C to 2°C and increasing to 3°C. This demonstrates the limits of pattern scaling properties and what is striking is that this is barely commented in the corresponding text of Sect. 3.1. Please at least add such comments on that, as pattern scaling is indeed one of the foundations of this manuscript. What is also missing here is the corresponding spatial patterns of such nonstationarity, which would be interesting for understanding the limits of pattern scaling across the UK.
- L.288: Fig. S1 is rather required in the main text in my view.
- Fig. 3: This is clearly unreadable as such, for several reasons. One, the color scale is non linear, which is definitely against perceptual rules (Hawkins, 2015 ; Stoelze and Stein, 2021). Two, maps are way too small. Three, using catchment surface as the support for colors is not appropriate for such small figure dimensions, as it perceptually highlights only large catchments. I would therefore strongly suggest using shapes at the outlet of each catchment. Four, colorscale label sizes are not homogeneous across facets, and the colorscale titles "location paramater" are redundant.
- Fig. 4: This is again much too small for being readable. And again, I would strongly recommend using e.g. disc shapes, and also using grey instead of black for UK costlines. A discrete color scale might also be more effective.
- Fig. 5: Please confirm (in the legend) that boxplots show differences across catchments only (and not across climate/hydrological models).
- L371-372: What can the reader refer to when reading that rarer droughts are accompanied by large variability? Across what? Spatial? Parameter estimation? Other?
- L421: Please provide factural elements to comfort the assertion of "robust estimates of uncertainty".
References
Hawkins, E. (2015) Graphics: Scrap rainbow colour scales. Nature, 519, 291. https://doi.org/10.1038/519291d
Sauquet et al. (2025), Évolution de l’hydrologie de surface en France par niveau de réchauffement, https://doi.org/10.57745/MN29RG, Recherche Data Gouv, V5.
Stoelzle, M. & Stein, L. (2021) Rainbow color map distorts and misleads research in hydrology -- guidance for better visualizations and science communication. Hydrology and Earth System Sciences, 25, 4549-4565. https://doi.org/10.5194/hess-25-4549-2021
Citation: https://doi.org/10.5194/egusphere-2025-4096-RC3 - AC3: 'Reply on RC3', Srinidhi Jha, 10 Dec 2025
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 14,870 | 1,167 | 210 | 16,247 | 517 | 313 | 293 |
- HTML: 14,870
- PDF: 1,167
- XML: 210
- Total: 16,247
- Supplement: 517
- BibTeX: 313
- EndNote: 293
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Lucy J. Barker
Jamie Hannaford
Maliko Tanguy
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(1674 KB) - Metadata XML
-
Supplement
(7802 KB) - BibTeX
- EndNote
- Final revised paper
Title: Evolution of nonstationary hydrological drought characteristics in the UK under warming
Recommendation: Accept after corrections
Explain the non-stationarity in the hydrological drought time series. How the future groundwater estimates are calculated and how accurate it is?
Line 35: 1.2 deg is the how many years average?
Line 80-81: ….”transient changes in low-flows characteristics”; what is the meaning of the transient changes here?
Line 113: What is “transient” is not clear from the introduction?
Line 117-118: “we aim to capture the full spectrum of possible future hydrological drought conditions under different climatic conditions.”
Line 133 – 135: “It should be noted that all 12 ensemble members originate from the same model framework and are based on the high emissions scenario (RCP8.5).” How it is same model framework? Please rephrase or write proper explanation for these lines?
Line 148: “recently developed CHESS-SCAPE” what do you mean by “recently developed”
Figure 1: which approach is more suitable for the drought identification; variable threshold or stationary approach? Have authors included the justification and applicability for these methods?
Conclusions:
Authors are advised to write the conclusions with the focus on the comparative analysis of the drought occurrence in the baseline and future periods under different warmings. Which can help in the framing/modifying the policy for the future dryness events.