the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 May 2026
Redistribution of frost weathering across Spain under recent climate warming
Abstract. Frost weathering (cryoclasty) is a key cryospheric process controlling rock breakdown, sediment production, and the mechanical evolution of landscapes at mid- to high latitudes. Its effectiveness depends on the interaction between temperature variability, moisture availability, and the frequency of freeze–thaw transitions. Here we analyse the spatial and temporal dynamics of frost activity across Spain during 1993–2022 using daily temperature and precipitation records from 84 meteorological stations spanning alpine, oceanic, inland Mediterranean, coastal, and subtropical environments. Five cryoclimatic indicators were evaluated to characterise both thermal severity and hydro-thermal effectiveness: Frost Days (FD), Freeze–Thaw Cycles (FTC), Intense Freeze–Thaw Cycles (IFTC), the Frost Intensity Index (FI), and the Wet-Frost Index (WFI).
Results reveal a strong concentration of effective frost weathering in mountainous and perimountainous regions, while frost activity is marginal or absent in lowland Mediterranean, coastal, and subtropical areas. Over the last three decades, most of Spain shows a statistically significant decline in frost days and freeze–thaw cycles, accompanied by a shortening of the frost season. In contrast, perimountainous belts exhibit sustained or locally increasing frost intensity and frequent transitions through the frost-cracking temperature window. The Wet-Frost Index highlights a progressive spatial contraction of moisture-effective frost conditions, with linear projections to 2050 suggesting that effective frost weathering will become increasingly restricted to high-altitude mountain environments. These findings show that climate warming is redistributing, rather than uniformly suppressing, frost-driven cryospheric processes at the southern margins of Europe.
- Preprint
(2214 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- RC1: 'Comment on egusphere-2026-1044', Anonymous Referee #1, 12 Jul 2026
-
RC2: 'Comment on egusphere-2026-1044', Riho Kido & Takuya Inoue (co-review team), 15 Jul 2026
General comments
This paper estimates the spatial and temporal variations in freeze-thaw weathering based on temperature and precipitation data by using climate data from across Spain. It uses multiple indicators to assess frost weathering and predicts changes in frost weathering from various perspectives.
Frost weathering is a phenomenon that causes changes in topography and the deterioration of historic structures. This study is important because it predicts how the areas where this process is active will change as temperature and precipitation patterns shift due to climate change. In particular, by focusing on changes in both temperature and precipitation at high resolution, the study enables predictions that accurately capture the influence of topography.
This manuscript addresses an important topic, and has the potential to make a valuable contribution to the field. However, further clarification is needed on several points in order to more fully support the main conclusions. In particular, the interpretation of the data needed to be explained more clearly, including the relationship between the frost weathering index used in this study and the freeze-thaw mechanism it represents. I therefore provide the following comments, which I hope will help the authors improve the clarity and strength of the manuscript.Specific comments
1) P6-L30
How did you reflect the aspect in the distinct geographic-climatic areas? In Figure 1, the scale of the distinct geographic-climatic areas looks much larger than the scale at which slope aspect can be reflected.2) P6-L23
It is necessary to explain more clearly which phenomena of frost weathering each indicator used in this study is intended to evaluate. For example, while the FTC and WFI use temperatures around 0°C or −1°C as thresholds, are these indicators based on the assumption of volumetric expansion caused by the freezing of water within the bedrock? On the other hand, in the case of cryoclasty caused by the ice segregation, a lower bedrock temperature range is considered important. Furthermore, since bedrock temperature does not equal air temperature due to the effects of solar radiation or thermal conduction, it is necessary to explain to what extent thresholds based on air temperature represent the freeze-thaw process within the bedrock.3) P7-L4
Regarding the definition of WFI, does it refer to precipitation without distinguishing between rainfall and snowfall? If there is heavy rainfall and the moisture content in the bedrock increases, freeze-thaw weathering is likely to increase; however, if there is heavy snowfall and a deep snow cover, the insulating effect of the snow cover should reduce freeze-thaw activity. If the WFI rises due to increased snowfall, I don't think frost weathering becomes more active.4) P9-L24
Why are there regions where FI is increasing? Was this upward trend in FI consistent from 1993 to 2022? Or was 1993 simply a warm year, and 2022 a cold year?5) P15-Table2
To indicate that temperatures are warming, wouldn’t it be better to show the trend in average temperatures rather than the extreme minimum temperatures before and after 1993? This paper should focus on long-term trends in frost weathering rather than on extreme events.
6) P17-L5
When we talk about the “stochastic nature of precipitation” in the context of climate change, doesn’t that mean an increase in extreme precipitation events? Or could it also mean a change where the number of days with no precipitation decreases, while the number of days with 2 mm/day of precipitation increases? In that case, the manuscript would benefit from a discussion of the types of regions where the number of days with no precipitation are expected to increase, as this would help readers better understand the complexity of precipitation changes.Technical corrections
1) In Figure 1, the colors for the hypsometry-specifically, the 900-1200 and 1200-2000 ranges are reversed, making it intuitively easier to understand because the brown becomes progressively darker as the hypsometry increases.2) There are no color bars in Figures 4 and 5, so please add them.
3) Looking at the bottom of Figure 4, it does not appear that the IFTC is declining sharply. Is it correct to understand that the IFTC continued to decline from 1993 to 2022? If so, please revise the figure so that the downward trend in the IFTC is clearly visible. For example, please revise it by drawing a straight line, as shown in Figure 6.
4) Isn't the 35.8°C on page 15, line 11 actually a typo for -35.8°C?
Citation: https://doi.org/10.5194/egusphere-2026-1044-RC2 -
RC3: 'Comment on egusphere-2026-1044', Anonymous Referee #3, 15 Jul 2026
This is an interesting topic for a paper given the current interest in the effect of climate change in reducing the number of frost weathering events in more temperate regions. The authors show the importance of regional scale analysis in examining frost weathering. I guess this need for high resolution is strongest with the wet frosts? I think the authors were right to examine a range of frost indices.
The paper is rather simple and descriptive and while there is nothing especially wrong with such an approach in work of this kind I worried that more was not made of the interpretation of what the results might mean in policy terms. Are there for example lessons for the management of landscapes or stone buildings under climate change? Especially issues such as the geographical distribution of such concerns. I realise the authors allude to this very briefly in the results section, but more at the end of the MS would be helpful.
Specific points
The title seems fine, but the keyword "frost weathering " might be replaced by one that is unique to the usage in the title, as that might enhance discoverability.
The paper adopts 4-sigma approach to detect outliers. I wondered about the normality of the data set. Was this checked with the Shapiro-Wilk Test or some equivalent.
I think the Mann–Kendall rank correlation test was an excellent choice for trends, I wondered if this would not make the Theil-Sen slope the correct choice for expressing the trend slopes. I realise that the authors have often expressed change as a percentage, but the Sen slope is perhaps more robust .
I was a little uncertain on the shifting time scales there is a 30-year time window (quite short to define a trend), but that changes later in the MS to cover more historical periods. Then towards the end, the MS is concerned with the future through to 2050. Perhaps more clarity on the choice of these timescales would help.
How reliable are the future projections. Although perhaps too much for the current project, would these not be better using CMIP ensemble and that could then be taken through to 2100 and examine various emission scenarios.
The future wet-frost frequency is seemed to be distinctive and I get the impression it is important in the Cantabrian mountains. I wondered what drove it to be important there.
Citation: https://doi.org/10.5194/egusphere-2026-1044-RC3 -
RC4: 'Comment on egusphere-2026-1044', Anonymous Referee #4, 17 Jul 2026
General comments:
The paper presents data on the spatial distribution and temporal progression of frost weathering in Spain, based on a 30-yr dataset of 84 meteorological stations evenly spread across the country. This is a comprehensive dataset with a good selection of meteorological stations. I like the “geographical” approach of using data representing diverse climatic regions. It is an important (and maybe sometimes overlooked) idea that certain frost weathering thresholds could shift geographically, especially when considering wetness. The paper is good in terms of quality: As far as I can tell, data standardisation, filling any data gaps, linear extrapolation etc. have been carried out carefully. The paper is well-written and almost free of typos.
Unfortunately, the paper has a number of fundamental weaknesses. The first and obvious is the use of air temperature data to derive frost weathering efficiency. It is well known that (a) air temperature and near-surface rock temperature fundamentally differ and (b) diurnal resolution is too low for deriving e.g. freezing rates. I well understand that for a country-wide study the desirable data is not available. However, the relation of air T to rock T is not even discussed, despite of the wealth of available literature.
Second, very weak temporal trends of single stations are over-interpreted and even projected 30 years into the future, with highly questionable results. Statistical significance parameters are not provided. In a nutshell: many results are self-understanding common climate change knowledge, and of the points that go beyond this some are questionable.
Third, the quality of the Discussion section is low. Organisation is poor: Section 4.1 is additional information which is not helpful; section 4.3 presents new data, which should be Results and not Discussion (but is also not helpful). The concepts of the frost-cracking window and of subcritical cracking are mentioned, but incorrectly explained; the interpretation is often speculative and not specifically related to their own results. See my numerous detailed comments.
For the given reasons the paper cannot be accepted in its present form. I recommend rejection with the option to be reconsidered after major revisions.
Specific comments:
P1 L10: Frost weathering efficiency depending on frequency of freeze-thaw transitions is overly simplifying and outdated. Consider the ample literature on the frost-cracking window.
P1 L12: Daily temperature is not very meaningful; meteorological stations (air temperature) are also not informative (or only for a very broad overview, which is OK as this is the intention of the paper).
P1 L17: What is “effective frost weathering”?
(All these points are mentioned in the Introduction buta reader who only reads the abstract will struggle with it.)
The first half of the Introduction is well-written.
P3 L1-6: “… this places Spain within a broader class of mid-latitude transition zones where warming is expected to redistribute frost-driven processes rather than simply eliminate them.” - This sentence is somewhat contradictory to “many mountainous areas in Spain lie close to the lower thermal limits of sustained frost activity” which sounds as if frost regimes might rather be eliminated.
P3 L9ff: Using maximum and minimum air temperature data is an outdated approach of characterising frost weathering – this should at least be mentioned and justified.
P3 L14: this is not “high-resolution climatic data”
Methods
P4 L7-24: The list of stations in the figure caption should be shifted to the Appendix; station altitude should be provided
In chapters 2.2 and 2.3 it is not fully clear to the reader for which purpose data processing and geospatial analysis were carried out (trend interpolation, Mann-Kendall rank correlation – why and for what?). You can more or less find it in the text, but it might be better to state the respective purpose at the beginning of chater 2 or the sub-chapters.
P7 L1-3: I wonder what significance the FI index (as described) is supposed to have – apart from the square root, it appears to be identical to (i) (FD)… do you sum up the square roots of the reached minimum temperature?
P7 L4-6: The WFI is a difficult one as it doesn’t cannot account for seasonal fluctuations, snow melt etc. However, I agree that this is probably all that can be extracted from station data. Still, the problem should be mentioned.
Results
High number of frost days in mountains is not very surprising…
P8 L5: elevation of La Molina station should be provided
P8 L6-8: The last two sentences of the paragraph are rather uninformative
P8 L13: “there is a clear delaying trend, with FF shifting from October to November or December” – do you mean, over the 30 years of observation?
P8 L20-22: “This indicates that spring is more prone to frost events than autumn, suggesting that the most effective frost-induced rock breakdown occurs between January and April/May.” This is quite an old school statement that could also be drawn from a table of monthly mean and minimum temperatures.
P8 L24/25: “the FFP can drop below 100 days in high-risk zones” – what defines a “high-risk zone”? The fact that mountain areas can experience frost almost all year round is not exactly new.
P9 L4/5: “These thresholds ensure the phase transition of interstitial water within the rock” – there is almost no threshold that “ensures” freezing of pore water. Better say: The lower the temperature threshold, the higher the probability that pore water actually freezes.
Fig. 4 is interesting, showing the deeper frost at the inland plateau areas compared to the higher Pyrenees.
Fig. 5: Interesting… but colour bars are missing and it doesn’t become clear how significant these changes are. From the lower graphs, the trend (and especially the difference between the stations) is not really convincing.
P11 L10: “The WFI (precipitation > 2 mm within 3 days prior to Tmn < -1°C) identifies the most effective frost events for rock weathering” - That’s far too strong a statement. By adding a very basic precipitation criterion to an air temperature threshold, the authors try to also take precipitation into account. Which combination of rock moisture, rock temperature, duration etc is most effective is an unsolved matter of ongoing research.
Fig. 6: The distribution of wet-frost index is not very surprising (mountain ranges with cool temperatures and high precipitation). However, the trend at most stations shown is very weak and almost horizontal (no change), except of station 40 which might be due to increasing T and decreasing P. Showing graphs of one of the “dark blue” Picos d’Europa or Pyrenean stations might be more convincing to visualise the contrast (increasing WFI).
Discussion
Section 4.1 is not Discussion; it is additional information on the study area. The connection to the data presented remains unclear.
P13 L9 ff: The argumentation in this paragraph is the wrong way round.
P13 L24 ff: “In high-alpine sectors, the prevalence of "ice days" may paradoxically limit this process by "locking" moisture in a frozen state.” The way this is written is a misconception. The frost-cracking window is somewhere between -3 to -6 C for high-porosity rocks and as low as -7 to -10 C for low-porosity rocks. Thus, ice days do NOT “lock” moisture in a frozen state. Sustained subzero temperatures between approx. -1 and -10 C allow slow migration of water in liquid an gaseous phase to the growing ice formations. A “frequent transition through the FCW” is not needed. Draebing et al. (2017) are correctly cited here: sustained frost events are required. Why “transitions through the frost-cracking window” might increase in “perimountainous zones” is hard to understand and not backed by the data.
P13 L34: “repeated cycling around 0 °C promotes subcritical crack growth and fatigue”: This is NOT the concept of subritical cracking. S.Cr. is a physico-chemical process that is mainly promoted by high pore water content. I suggest to read the ground-breaking papers of Eppes et al. before using the term. What is more, transitions through the (undefinded) frost-cracking window and “repeated cycling around 0 C” are all thrown together here.
P13 L37: It is not clear from the presented data why there should be “short but intense freeze-thaw periods” in the future.
P14 L1-5 is pure speculation which is not at all backed by the data.
Chapter 4.3 presents new data, which should be Results and not Discussion. What is more, this data is neither new nor helpful as it shows the well known temperature shift due to global warming (as the authors state on P15 L1-2). The paragraph P15 L9-12 is absolutely meaningless in the context of the paper.
Section 4.4 Future Projection: Projecting the, in parts, very weak trends of wet-frost index further 30 years into the future is courageous. The resulting blue blob over Picos d’Europa (Fig. 7 lower right) looks spectacular but is not at all backed by any data.
P17 L4-6: I give the authors credit for recognising the problem themselves: “However, long-term projections for the WFI must be interpreted with caution. When focusing on the most recent decade (since 2012), the trend directionality becomes erratic, highlighting the stochastic nature of precipitation in the Mediterranean context.” I fully agree, which is why the respective trends should not be shown (Fig. 6, Fig. 7).
Conclusions: These are heavily impaired by the mentioned shortcomings in the Results section and thus, are not commented here any further.
Editorial comments
P2 L4: “it also plays”
P13 L16: Properly cite Noah’s Ark project
Citation: https://doi.org/10.5194/egusphere-2026-1044-RC4
Data sets
Climatic and cryoclimatic indices, historical trends (1993-2022) and 2050 projections Javier Martínez-Martínez, Carlos Gabriel Morales, María Teresa Ortega https://doi.org/10.5281/zenodo.18851330
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 83 | 43 | 9 | 135 | 8 | 9 |
- HTML: 83
- PDF: 43
- XML: 9
- Total: 135
- BibTeX: 8
- EndNote: 9
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
The manuscript entitled "Redistribution of frost weathering across Spain under recent climate warming" by Morales, Ortega and Martinez, 2026, examines the spatio-temporal dynamics of frost activity across Spain during 1993-2022, using daily temperature and precipitation records from 84 meteorological stations. The authors define five cryoclimatic indicators to characterise thermal severity and hydro-thermal effectiveness, using physiographic and hypsometric criteria to split the station dataset into five environment groups. The authors apply quality controls to correct bias in the station dataset, employs geospatial analysis to evaluate the spatial variability of freezing mechanisms across the Spain territory, and provides future projections for two of the five cryoclimatic indicator groups.
I find the subject matter interesting and the application to Spain's diverse climate regions could offer valuable insights. The influence of cryoclastic dynamics is unquestionably significant, and the cartography that the authors use the most to present the result is well-structure. The study's concept and structure could be replicable and extrapolated to other nations with a similar well-established database. However, the article’s structure is not yet appropriate and presents ambiguity regarding what the authors can obtain from the analysis, tending toward inferred analyses rather than direct findings.
The results and discussion are mainly descriptive and based only on "representative" stations, with minimal use of statistical metrics or state-of-the-art reference sources. Additionally, the document lacks tables to illustrate a general picture of the total analysis content, making it difficult for readers to fully engage with the descriptions and characteristics of the results obtained.
The authors´ primary focus on describing and inferring major discussion points, such as presenting synoptic drivers and atmospheric dynamics as the first picture of the discussion, makes this more of a descriptive essay than a proper scientific discussion of what the authors have found through their analysis.
Suggestions for the Authors - Major Concerns
Before submitting it again, I suggest the authors: (a) clearly state their hypothesis and research questions in the introduction; (b) present a comprehensive statistical summary of all stations; (c) justify the selection of “representative” stations with quantitative criteria; (d) move methodological details currently in discussion to methods section; and (e) reorganise results to present all stations’ data before focusing on representative examples only. Additionally, I highly recommend to check and improve typos and English structure that would make the text’s writing more appropriate.
Line-by-Line Comments
Abstract
Lines 19 and 22: Use acronyms consistently or skip them entirely. E.g., Wet-Frost Index (WFI) – define at first use.
Introduction
Line 15: “Air temperature" is sufficient within geoscience approach.
Lines 26–28: This general idea could be improved with a literature reference.
Page 3, Lines 2–8: Many general affirmations with no references.
Page 3, Line 11: "Climate aggressiveness" sounds unusual. Consider "the impact of climate on (...)".
Page 3, Lines 9–16: These lines do not properly reflect the specific goals and hypotheses. Please rephrase this paragraph to be more specific and clearly state the goals.
Methodology
Page 4, Lines 7–24: Provide this data differently: a table would work better. Provide more information on each station (latitude, aspect, etc.), to help readers interpret your network selection design. Consider moving to supplementary material.
Statistical procedures: Were any supported by references? Is this a novel and ever unique approach? Many similar studies have applied similar criteria and should be cited.
Page 5, Lines 8–22: The data processing and quality control mentioned here could be included in the suggested table. As written, it is ambiguous. Was it implemented for all stations or only those in mountain climates or semi-arid Canary climates?
Page 6: Which ArcGIS version? Which package and exact function? Which grid size was used?
Page 9, Line 17: "High altitude" typo.
Page 9, Line 18: This belongs in the discussion. If presenting novel results, why support with a reference?
Figures 5–6 caption: "Average values of the frost intensity parameter by geographic distribution". Clarify it please.
Page 11, Lines 15–17: Move to discussion better
Figures 2–6: The selection of representative stations is not supported. Approximately 80 stations were used, yet only 8–5 stations are presented consecutively. Why were these specific stations chosen?
Page 8, Line 7: This is inferred from index results but not linked to the method. To what really extent these frost conditions truly affect infrastructure in cities, airports, or surrounding areas where the stations are located?
Discussion
Section 4.1: While radiative frost (anticyclonic ridge) and advective frost (Arctic/polar air masses) are correctly identified at the regional scale. If this approach is maintained, you must also mention other local factors such as altitude and aspect that govern frost occurrence, particularly at many of the weather stations used.
Page 13, Lines 10–11: Already mentioned in the introduction.
Page 13, Line 36: Was this fact obtained from your analysis? Otherwise, it should be cited.
Page 14, Section 4.3: This should be moved to methodology and results.
Table 14: Again, Why are “representative” different stations used for this analysis?
Page 15, Line 6: "Absolute minima" – clarify.
Section 4.4: Again, this should be presented earlier in methods and results rather than appearing unexpectedly. By the way, how were predicted maps obtained?
Page 17, Lines 21–32: This appears to present a limitation of the study. It would be more effective to separate this into a distinct category to clarify the point.
Conclusions
Page 18, Lines 1–2: Better to present a table with these results than to comment randomly throughout the manuscript.
Page 18, Lines 19–24: This is a sharp conclusion but not well-supported by the presented evidence. With only a few maps and tables with few stations, this statement cannot be sufficiently supported.