The humidity paradox: how drier conditions and fewer clouds amplify terrestrial warming due to global climate change

Glantz, Paul; Noone, Kevin James; Devasthale, Abhay; Schenk, Frederik

doi:10.5194/egusphere-2026-1654

Preprints

https://doi.org/10.5194/egusphere-2026-1654

Preprints

02 Apr 2026

| 02 Apr 2026

Status: this preprint is open for discussion and under review for Earth System Dynamics (ESD).

The humidity paradox: how drier conditions and fewer clouds amplify terrestrial warming due to global climate change

Paul Glantz, Kevin James Noone, Abhay Devasthale, and Frederik Schenk

Abstract. Clouds are a controlling factor determining the planetary albedo and thus strongly contribute to establishing the radiation balance at the top of the atmosphere as well as the energy balance at the surface. Terrestrial warming rates, divergent to oceans, have accelerated substantially since around 1980, along with significant changes in humidity and cloud cover. We analyse spatiotemporal changes in warming rates compared to changes in Earth’s radiation and humidity, considering land globally for the period 1979–2023, using reanalysis and satellite data. We find statistically significant increases in top of the atmosphere net solar radiation and surface net terrestrial radiation, causing a net warming, together with decreasing cloud cover. These changes coincide with drier land surface conditions and are associated with the humidity paradox: insufficient supply of water vapour causing a decrease in relative humidity. Reduced evaporative cooling of the land surface is an additional positive feedback and has likely contributed to the ocean-land warming contrast. The oceans have, on the other hand, effectively unlimited water to evaporate and can therefore cool in a warming climate by evaporating more and more water. Inhibited cloud formation over land and an increase in solar radiation provide an amplifying feedback loop for the observed rapid terrestrial warming in recent decades due to a CO₂-driven humidity deficit. A decrease in precipitation over land regionally is a strong indication of perturbed surface water balance that is driven by increases in absorbed infrared and solar radiation.

Received: 24 Mar 2026 – Discussion started: 02 Apr 2026

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.

Download & links

Paul Glantz, Kevin James Noone, Abhay Devasthale, and Frederik Schenk

Status: open (until 25 Jun 2026)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2026-1654', Anonymous Referee #1, 20 May 2026 reply
The study by Paul Glantz et al. addresses an interesting and scientifically sounded idea, namely, it looks at the effect of differential skin surface temperature trends over land and the ocean. In theory, slower warming of the ocean surface should limit water vapor supply to the atmosphere over land, and thus, cause reduction of atmospheric relative humidity and cloudiness. In practice, the atmospheric hydrological cycle is very complex, mediated by numerous poorly understood feedbacks and even less understood atmospheric dynamics, especially in the turbulent planetary boundary layer. The study does not attempt to penetrate atmospheric physics or dynamics. It focuses on a broad search for statistical evidence of terrestrial warming amplification through reduced moisture and cloudiness. If properly implemented, the study could be an important contribution to physical climatology and climate change science. Unfortunately, in my opinion, the study failed in addressing its scientific goals. The reported results are messy and do not warrant publication.
The major of the study is that it does not apply proper statistical instruments for the analysis. It rather loosely browses across different statistics derived from different diverse datasets. We find here model-based atmospheric reanalysis (ERA5), satellite products of different quality, resolution, and coverage (CLARA, CERES), in situ datasets (HadISDN, GPCP, GPCC). What does this zoo of datasets make to the question in scope? It makes the text long, logically fractured, and complicated to read. I don’t understand the needs for such complications. Just opposite, in my opinion, if ERA5 – as an internally consistent and complete dataset – demonstrates the constraining effects of differential warming on the hydrological cycle, it should be sufficient to analyze ERA5, and then, in Discussion, to show how good/bad the reanalysis is with respect to observational datasets (and whether it does/doesn’t matter).
So, the most important part of any study – the method – is actually missing in the Data and Method section. This is for a good reason though because the authors do not have any logically justified analysis method for their study, but just a more or less trivial set of popular statistical indicators, mostly trends, of more or less relevant variables. The presented set of statistics does not support or question the main hypothesis of this study.
My other comments are less significant but still pinpoint serios weaknesses and problems with the manuscript.
The manuscript is poorly written. Many sentences and even paragraphs are impossible to understand. An unconventional terminology is used in several places. E.g., what is “the humidity paradox”? I do not see any “paradox” in the discussed effect. What is “climatology scene”? E.g., line 180, “data product … mounted on the Terra and Aqua spacecraft”. How could a data product be mounted on a satellite? How CERES could simultaneously be a data product, a project, and an instrument? E.g., line 200, “The irradiances are calculated based on temperature …” Are irradiances measured or calculated? How could cloud fraction be corrected “in the same way”, based on temperature?

Physical interpretations are frequently unclear or directly incorrect in the manuscript. E.g., around line 35, “The specific heat capacity is fairly insensitive … and cannot explain amplified land warming”, just opposite, it is sensitive as follows from the heat balance equation, and it can partially explain the amplified warming, and that it exactly what S. Manabe showed and many other studies confirmed after him. g., around line 45, “Moist air … has a higher heat capacity and resists changes in temperature”. This is true, but this is minor effect; the major effect is that water vapor is the greenhouse gas by itself.

Discussion of cloud effects is very shallow and problematic. Different types and layers of clouds have different effects and feedbacks. Those are not discussed. Also the important fact that convective clouds cover only a small fraction of the ERA5 grid cell and therefore have only a limited effect on the radiation budget, has not been discussed.

Statistical analysis (supposed to be in section 2.7) should be central for the statistical study, and hence, it must be carefully described and justified, but it wasn’t. I found only a few trivial equations for some statistical indicators. As minimum, the following are required:
How were different datasets homogenized to provide comparable statistical quantities? I see that they were not as relative humidity at 1000 mb in ERA5 is compared with the surface values in HadISDN (Figure 3)

How were different spatial/time resolution of datasets handled?

How were trends calculated? How were data outliers processed?

How were smoothing and filtering applied?

All dataset descriptions are rather poor and unclear. Section 2.5 is not a description but rather declarative advertisement of GPCP. The description must be homogeneous and focused on the features relevant to the study.

Figure 9 is supposed to be important, but I could not understand what it shows.

Figure 10 is rather obscure to me. Does it show that since 1979 the amount of reflected solar radiation has increased by 100% almost everywhere?

Figure 13 should be Figure 1 as it supposedly shows the effect (differential warming) which this study addresses.

Discussion in lines 660-675 is very confusing. It touches the Amazon and takes results from boreal areas. I did not understand the conclusion here. Is it less absorbed radiation – stronger warming?

Reply
Citation: https://doi.org/10.5194/egusphere-2026-1654-RC1
- AC1: 'Reply on RC1', Paul Glantz, 16 Jun 2026 reply
  
  Publisher’s note: the supplement to this comment was edited on 22 June 2026. The adjustments were minor without effect on the scientific meaning.
  
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-1654-AC1
RC2:
'Comment on egusphere-2026-1654', Anonymous Referee #2, 07 Jun 2026 reply

The humidity paradox: how drier conditions and fewer clouds amplify terrestrial warming due to global climate change, by Glantz et al
As the title suggests, this study aims to argue for an important role for drier conditions (in a relative humidity sense) and associated reductions in clouds to contribute to the relatively greater increase in warming over land compared to ocean. It does this by showing long term trend in a wide variety of surface energy and water balance related variables from ERA5 reanalysis and other datasets. Unfortunately, I think this study falls considerably short of providing a coherent, quantitative story to prove this point and instead reads more like a laundry list of trend analyses from these datasets, many of which have already been presented in prior work. I, therefore, do not think this paper meets the required standards for publication.
(1) My primary criticism is the lack of quantitative analysis to prove the connections that the authors are arguing for. Trends are shown in many quantities, but many of these trends are already well established in the literature e.g.,
Humidity trends: Dunn et al. (2017) cited in the paper, Douville and Plazotta(2017) cited in the paper, Simpson et al. (2024) PNAS https://doi.org/10.1073/pnas.2302480120 not cited in the paper
Top of atmosphere radiation trends: The Loeb et al studies cited in the paper
Downward shortwave trends at the surface: Wild et al. (2026) https://doi.org/10.1007/s00376-025-4534-2, McKinnon and Simpson (2025) https://doi.org/10.1029/2025GL119493
The paper is extremely descriptive with a lack of novel analysis or discussion to bring together the trends that are shown into a coherent story with a quantitative demonstration of the claim made in the title that reduced clouds associated with drier conditions are an important contributor to the amplified warming over land compared to ocean.
(2) ERA5 trends in precipitation, sensible and latent heat and radiative fluxes are presented without any acknowledgement that they are heavily influenced by the representation of processes in the underlying model.
(3) Some statements are made about the reductions in atmospheric humidity being related to the lack of water vapor being transported in from the oceans because they haven't warmed as much. I don't think any evidence has really been presented to prove that this is the dominant factor as opposed to e.g., reduced evapotranspiration from the land surface.

Comments by line number:
The title, l19, and elsewhere: It's not clear what is paradoxical about this. I wouldn't call it a paradox. It is simply that there is something that is limiting the water availability, whether it be evapotranspiration from the land surface or transport from elsewhere and as a result, atmospheric humidity is not rising at the rate required to maintain constant relative humidity in a warming atmosphere.
l64: I think there are some other relevant references for this point e.g., Vautard et al. (2023) https://doi.org/10.1038/s41467-023-42143-3 and Singh et al. (2023) https://doi.org/10.1038/s43247-023-01096-7
l130: I don't think there is any rule that says that the water balance of the land surface nees to be maintained in a transient climate
l141: "The latter" doesn't sound right in this context. I'm not sure what it's referring to.
l295: There are also fairly substantial differences around Antarctica.
Figure 3: Why use 1000 hPa relative humidity? This is going to involve some extrapolation below the surface. 2m surface air temperature and dew point temperature can be used which would be more comparable to HadISDH.
l478: The decline in specific humidity in the southwestern USA has already been extensively discussed in the literature (see studies mentioned above)
Figure 2: There are really big differences in terrestrial upward radiation between the products over Africa, as well as relative humidity in Figure 3 and then Figure 11 makes it clear that ERA5 has spurious precipitation trends in this region compared to GPCP, so likely can't be trusted. No attempt has been made to link these different features together and it raises the question of where and when ERA5 can be trusted for certain fields.
l710/711: This sentence makes it sound like natural variability goes away with time. Of course it doesn't, but its incontribution to long term trends reduces.
l722: Not sure the word "intimate" is appropriate here, but I'm not sure what is meant.

Reply

Citation: https://doi.org/10.5194/egusphere-2026-1654-RC2
- AC2: 'Reply on RC2', Paul Glantz, 18 Jun 2026 reply
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-1654/egusphere-2026-1654-AC2-supplement.pdf
  Reply
  
  Citation: https://doi.org/10.5194/egusphere-2026-1654-AC2
RC3:
'Comment on egusphere-2026-1654', Anonymous Referee #3, 25 Jun 2026 reply
Summary:
The paper investigates recent observed trends in the land’s climate and proposes a positive feedback between land humidity and cloud cover. While the research topic fits within the scope of ESD and is definitely worth pursuing, the presented analysis unfortunately lacks in clarity, and does not manage to show physical causation associated to the presented correlations. The central line of argument is hidden in a cross-comparison of datasets and variables with unnecessary figures, instead of presenting a quantitative analysis to physically support the proposed feedback. It is evident that there is reduced land cloud cover over land and that this is expected to lead to a positive forcing on land warming due to more short wave downward radiation, however this result was already presented in Loeb et. Al (2021). The correlations presented in the subitted manuscript are not sufficient to justify the author's claims that 1) reduced cloudiness significantly amplified land warming, because the expected warming is not quantified, and therefore 2) trends in land humidity and cloudiness signal a positive feedback, and not e.g. a common cause. Therefore it is not advisable that the manuscript is accepted as submitted. Further analyses are required to prove and quantify the expected magnitude of the feedback. Considering all above, the authors are strongly encouraged to contiue pursuing this research question, but further analyses and a full rewrite of the manuscript are advised.
Major scientific comments:
The use of the term paradox does not seem fitting, or at least does not become clear. In the abstract is seems like decreasing relative humidity and cloudiness are supposed to contradict each other, and are later also described as feedback, but they do not. In line 717 in the conclusion, it is said that the fact that land humidity does not increase at the same rate as the saturation vapour pressure is a paradox, but this is not true either. What is the scientific gap the authors are trying to understand? In line 135 it is called a positive feedback between drying and shortwave radiation which is very plausible and seems to be the subject of research, but this is not reflected by the title and abstract and conclusion, which discuss a paradox.

The conclusions (line 725) read like the the proposed feedback between humidity and cloud cover/shortwave radiation is claimed to explain enhanced land warming, not only an amplifying feedback. However, there are existing explanations for enhanced land warming, like the moisture-convergence constraint leading to a higher temperature lapse rate in the boundary layer by Byrne&O’Gorman (2018), mentioned in the manuscript. Please clarify if you understand the feedback as separate mechanism or part of Byrne&O’Gorman’s mechanism.

The manuscript is missing a quantitative link between the proposed increase in land surface SW radiation and the expected land warming. How much of the enhanced land warming is caused by the effect described by Byrne&O’Gorman and how much could the proposed feedback add? What is the cloud radiative effect over land? How much does the clear sky Planck response compensate? From the manuscript it does not become clear if there is a significant effect of the cloud reduction, or if land warming is not completely dominated by the effect proposed by Byrne&O’Gorman and competing effects like co2 increase (higher forcing on the dryer land), or aerosol cleanup.

Previous studies have shown that the stability over land is strongly influenced by the mean SST warming level (see Chadwick 2019). In AMIP experiments with isolated uniform SST increase, land moisture convergence decrease and stability increase. The presented correlations do not distinguish whether declining humidity causes reduced cloudiness, whether reduced cloudiness causes declining humidity, or whether both are responses to large-scale warming and circulation changes. Can you refute a SST increase as common cause of the two trends?

Why are the South American continent and the northern mid/high latitudes chosen for analysis if the statements are made for the global land? The chosen domains seem to be too broad to identify distinct regimes but still miss out on significant parts of global land. Would e.g. a zonal mean analysis not be a better suited approach to separate regimes?

It is does not become clear why both reanalysis and observations are analysed. Since ERA5 has radiation simulated by IFS without assimilating any observations (Hersbach et al., 2020). Why is not only CERES used for the radiation trends, if ERA5 features radiation simulated by IFS that is not assimilated? It would be easier to follow if the feedback would be shown on a minimal set of datasets and variables, and in an optional second step it would be checked if a possible feedback is apparent in other sources.

Further comments:
Introduction
This section lacks a through line. Further, the research gap is not clearly stated.
Line 91: „Clouds affect outgoing longwave radiation in the same way as GHGs“ - This statement is not correct, even with the conditional clause that follows, because their effect depends on altitude and is spectrally much more uniform. Saying they operate “in the same way” is physically misleading.
Methods
This section seems a bit overly detailed, specifics of the products that are not relevant to this study can be looked up in the release papers of the products.
Line 165: It is mentioned that ERA5 temperature is used, but it is inconsistent which variable (skin layer, 2m). Why is skin temperature not used throughout? Like for humidity, there is opportunity to make the paper more consistent, and introduce the specific variables here.
Line 259: Why does a low spatial resolution prevent area-weighing?
Results
The discussion of humidity changes is partly unclear. It’s comprehensible that relative humidity is a 0th order measure for cloud formation because it directly encapsulates how far air is from condensation, and constant relative humidity still is a common null hypothesis. However, the analysis of changes in specific humidity (Figure 4) and the ratio of column water vapour over 2m temperature doesn’t seem to be relevent for the argument. Further, it is unclear why the latter quantity would be a meaningful quantity when there are accepted measures like the column relative humidity.
Subsection 3.6: It’s doesn’t become clear why precipitation is taken as indicator for land humidity and not moisture convergence (or mean P-E as analog). It seems like the statement in line 23 could be made if P-E had ben analysed, but not only analysing P. Changes in precipitation and evaporation can balance each other in terms of the moisture budget and be more dependent on changes in stability and vertical motion than humidity. Also, an increase in shortwave radiation is often associated with an increase in precipitation (e.g. Wild et. al. 2018), which opposes your argument. How do you reconcile this, and why do you use precipitation instead of P-E?
Line 690: An increase in the Bowen ratio could also be caused by a decrease in the sensible heat flux
Line 704: This is what is expected from Byrne&O’Gorman but it reads like it was a new finding of this study
Figures/Tables:
Generally it’d be helpful if regions of non-significant trends would contain hatches. Some figures don’t seem necessary for the line of argument and could be excluded or put in a supplementary section (Figure 4,9,10,11,12).
Table 1: The table is much too large and difficult to interpret. It seems like the data would be perfectly suited to create a boxplot or violinplot, please consider such a visualization style instead of a table. Further, the angle brackets defined in the caption could be used in Eq. 1. This could also be considered for Table 2.
In Figure 6, a diverging colorbar would immensely help identifying the sign of change. Again, hatching would be helpful to denote significance.
In Figure 10, it is unclear what ‚100%‘ refer to. Judging from the colorbar label it seems like all yellow regions show a 100% increase, when it probably means 0 change. Further, it seems like seasonal means would be sufficient, and for consistency with the other figures it would be nice to show the trend per decade.
Figure 11 It would be clearer if the period (yr) was in the colorbar label
Issues of understanding/precision:
Line 722: „crossing of the vapor-liquid phase“: Do you just mean to say evaporation?
Line 665: „Luo et al. (2024b) and Næss et al. (2025) found a reduction in the clouds due to deforestation“ Considering the many metrics for clouds, which one are you referring to?
Line 667: „The overall biophysical effect of deforestation“: please specify which effect you are referring to.
Line 688: „natural level“ Do you mean preindustrial?
References:
M.P. Byrne, & P.A. O’Gorman, Trends in continental temperature and humidity directly linked to ocean warming, Proc. Natl. Acad. Sci. U.S.A. 115 (19) 4863-4868, https://doi.org/10.1073/pnas.1722312115 (2018).
Chadwick, R., Ackerley, D., Ogura, T., & Dommenget, D. (2019). Separating the influences of land warming, the direct CO2 effect, the plant physiological effect, and SST warming on regional precipitation changes. Journal of Geophysical Research: Atmospheres, 124, 624–640. https://doi.org/10.1029/2018JD029423
Hersbach H, Bell B, Berrisford P, et al. The ERA5 global reanalysis. Q J R Meteorol Soc. 2020;146:1999–2049. https://doi.org/10.1002/qj.3803
Loeb, N.G., Ham, SH., Allan, R.P. et al. Observational Assessment of Changes in Earth’s Energy Imbalance Since 2000. Surv Geophys 45, 1757–1783 (2024). https://doi.org/10.1007/s10712-024-09838-8
Martin Wild et al. ,From Dimming to Brightening: Decadal Changes in Solar Radiation at Earth's Surface.Science308,847-850(2005).DOI:10.1126/science.1103215

Reply
Citation: https://doi.org/10.5194/egusphere-2026-1654-RC3

Paul Glantz, Kevin James Noone, Abhay Devasthale, and Frederik Schenk

Viewed

Total article views: 508 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
358	120	30	508	33	26

HTML: 358
PDF: 120
XML: 30
Total: 508
BibTeX: 33
EndNote: 26

Views and downloads (calculated since 02 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	229	70	21	320
May 2026	105	36	8	149
Jun 2026	24	14	1	39

Cumulative views and downloads (calculated since 02 Apr 2026)

Month	HTML	PDF	XML	Total
Apr 2026	229	70	21	320
May 2026	105	36	8	149
Jun 2026	24	14	1	39

Viewed (geographical distribution)

Total article views: 507 (including HTML, PDF, and XML) Thereof 507 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 25 Jun 2026

Short summary

Terrestrial warming rates, divergent to oceans, have accelerated substantially since 1980. Decreases in relative humidity, cloud cover and precipitation in addition to dryer land conditions is a strong indication of perturbed surface water balance that is driven by increases in absorbed infrared and solar radiation at the land surface. An amplifying feedback loop involving humidity, clouds, and solar radiation has likely contributed to the rapid terrestrial warming under global climate change.


Total:	0
HTML:	0
PDF:	0
XML:	0