Droughts in western Central Europe and associated atmospheric circulation patterns since 1844

Neimry, Emile; Goosse, Hugues; Jonard, Mathieu

doi:10.5194/egusphere-2025-5490

Preprints

https://doi.org/10.5194/egusphere-2025-5490

Preprints

02 Dec 2025

| 02 Dec 2025

Status: this preprint is open for discussion and under review for Weather and Climate Dynamics (WCD).

Droughts in western Central Europe and associated atmospheric circulation patterns since 1844

Emile Neimry, Hugues Goosse, and Mathieu Jonard

Abstract. Droughts in western Central Europe have major impacts on agriculture, ecosystems, and society, yet their long-term variability and drivers remain poorly understood. This study investigates drought variability over the past 180 years and its link to atmospheric circulation to identify their dynamic drivers. Three reanalysis datasets (ERA5, 20CRv3, and ModE-RA) are used to detect meteorological drought events via the 3-month Standardized Precipitation (Evapotranspiration) Indexes (SPI-3 & SPEI-3) and to connect them to atmospheric circulation patterns through k-means clustering. Dataset reliability is assessed over western Central Europe, providing consistent coverage from 70 to 165 years. Results show that recent severe and successive droughts, such as the 2018-drought, have historical precedents and display strong multidecadal variability. Diverging trends between SPI-3 and SPEI-3 over the last decades indicate an increasing role of atmospheric evaporative demand (AED). Summer dryness has intensified over the past 180 years, whereas winter dryness has declined. Regional and seasonal contrasts further emphasize the complexity of drought dynamics. Four distinct circulation patterns associated with droughts are identified: the Baltic High, British Isles High, North–South Dipole, and European High. Over time, droughts have become increasingly linked to the European High, a pattern characterized by strong AED anomalies, intense droughts, and this had a central role in the recent spring drying. The findings highlight the recent emergence of circulation patterns that enhance AED, marking a shift in the dynamic drivers of regional droughts under climate change.

Received: 05 Nov 2025 – Discussion started: 02 Dec 2025

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

Preprint (PDF, 7800 KB)

Supplement (16075 KB)

Download & links

Preprint (7800 KB)
Metadata XML
Supplement (16075 KB)
BibTeX
EndNote

Emile Neimry, Hugues Goosse, and Mathieu Jonard

Status: open (extended)

Post a comment Subscribe to comment alert

RC1:
'Comment on egusphere-2025-5490', Anonymous Referee #1, 03 Feb 2026 reply
Neimry et al. have analysed historical drought trends in western Central Europe (BE,NL,LUX,partFR,partDE). Using three different reanalysis products and four meteorological stations, they calculate SPI3 and SPEI3 and analyse trends and circulation associated with drought events. I find parts of the study interesting, for example the circulation drivers, the long historical perspective and the decadal variability in drought occurrence shown in Figure 5. However, I sometimes got lost in reading the results sections where many statistics are noted and discussed, often in a very lengthy way.
My first point of advise to the authors is thus to make an attempt at shortening the manuscript, by being more direct or where possible leaving out some explanations or moving things to the supplementary materials. An example, at the start of your data section:
“In addition to the identification of droughts, an objective is to connect them to atmospheric circulation. Consequently, reanalyses are pivotal datasets for this study. A reanalysis is a combination of historical weather observations and weather forecasting models, using a method called data assimilation (Slivinski et al., 2019; Valler et al., 2024). This approach provides a comprehensive dataset that represents fields consistent with each other through the application of physical laws within the model and the elimination of spatial and temporal gaps.”
In my opinion this whole bit can go. The objectives are already listed in the introduction, and I think the readers of WCD can be expected to know what reanalyses datasets are. Hopefully by critically assessing the text as a whole, you will be able to achieve a more streamlined and clear manuscript, that better showcases your work and results.
My second point is a hesitation regarding the fact that all analysis (outside Fig 8a/b) is related to SPI or SPEI trends/values, rather than the actual physical input variables of precipitation, temperature and evapotranspiration. I struggle with terms like ‘dryness’, when SPI/SPEI indicate a relative position on the wet/dry spectrum. Please consider performing or adding (in the SI?) Some analysis on the physical variables, and using the language to go with that (related to my point on clarity before). E.g. I believe that when you write about a ‘decline in dryness’ measured by SPI, you in fact mean an increasing precipitation trend. As a physically trained climate scientist that is easier to interpret.
Finally, I miss a critical evaluation in the discussion of the validity of your results and assumptions. Throughout the results section there are quite a few places where the three reanalysis products thoroughly disagree with each other, even when a common period is used. No statements or conclusions are made regarding this, whilst for me this is very important. Furthermore, given your interesting Fig. 5, should you not comment more on the influence of interannual/multidecadal variability on trends in relation to climate change trends (signal to noise discussion)? In a similar matter, I miss a critical discussion on the use of Thornthwaite, which as far as I know is a dated method of calculation PET. I fully understand that when only monthly data is available, this is your only option, but this will influence your computed trends and results, so I expect a critical discussion of implications.
As such, my advise to the editor is to offer the authors a chance to reword, and consider the manuscript again after major revisions have been made.
Major comments
I’m based in the Netherlands, a prominent part of your region of interest. I’m most familiar with the Dutch literature on drought, and these are often missing in your reference list. I list some, as a suggestion to extend your literature review, note this is not a complete list. These are a mix of trend analyses, event analyses and future projections: Philip et al. (2020), Zscheischler et al. (2020), Huang et al. (2021), Van der Wiel et al. (2021), Mens et al. (2022), Aalbers et al. (2023).

Line 157: I’m confused how one can reliably add detail to datasets by interpolation, as you do by upscaling from 1.8 degree resolution to 0.25 degree resolution. Have you verified what the resulting (spatial) fields look like?

Why have you chosen to use ERA5-land rather than normal ERA5? You don’t use any of the land surface variables in your analysis, that would benefit from the additional offline land-model calculations?

Line 306: In your analysis of the quality of the different reanalyses datasets, you fully rely on SPI/SPEI variables which are normalised, derived variables. Please check the reliability of input variables as well, e.g. PR, TAS, PET. Similarly, I think your analyses of trends may be helped by analyses of the input variables rather than SPI/SPEI alone.

Line 205 and 327: Thornthwaite, through its solo dependence on temperature, likely shows a stronger (too strong) trend over the analysis period towards higher PET values. This can somewhat be seen in Figure S1, where the red line has a stronger slope towards low SPEI values (drought). Similarly, in line 327, SPEI-3 drought occur at the end of the time series, whilst SPI-3 droughts occur more at the beginning. Can you analyse, maybe in a similar figure to S1, how the PET differences are, i.e. without the step towards SPEI? Please then also add a point in the discussion on this, and how you think the choice for Thornthwaite has impacted your results (section 7.4 I guess).

Line 403: what does SPI3 minus SPEI3 physically mean, it is a difference between normalised quantities that partially rely on the same input variable. Wouldn’t it make more sense to asses the trend in (normalised if you wish) PET, rather than in this strange quantity?

Section 6.2: This section is very dense with numbers and comparisons, but as a result difficult to follow and distill your main results from. There are a lot of differences mentioned, which are partially attributed to differences (biases?) in reanalysis datasets, but without coming to a conclusion. The influence of multidecadal variability is not mentioned, even though we know (and you showed in Figure 5) that this is of large influence especially for precipitation.

Minor comments:
Line 21: “raising particular concern” while I agree there is concern, I’m not sure it is more concerning in this region then in other regions globally. The Mediterranean, or Sahel, see larger trends towards droughts, transitioning to seasons without significant rainfall. The trends in Central Europe are relatively smaller, the climates relatively wetter and cooler.

Line 45: the transition zone between drier South and wetter North in climate projections is also a prominent feature of the Netherlands national climate scenarios (Van der Wiel et al. 2024).

Line 67: You only use CHDs six times in the manuscript, do you need the abbreviation?

Line 120/Figure 1: Generally I’d advise to avoid using reanalysis-based precipitation products, can you recalculate these values using (for example) the E-OBS gridded observational data? Do mention this limitation in your discussion as well.

Line 157: Please add details on your method to bias adjust to the 1979-2008 climatology, do you remove all trend or calculate the bias in this period and adjust the full dataset according to this bias?

Line 191: I think you have calculated SPI/SPEI at grid point level, and then spatially averaged, but am not 100% sure due to “In order to obtain the SPI-3 and SPEI-3 values for the entire western Central Europe region, precipitation and PET values for all land cells have been averaged over the study area.”, can you clarify?

Line 205: Thornthwaite, through its solo dependence on temperature, likely shows a stronger (too strong) trend over the analysis period towards higher PET values. This can somewhat be seen in Figure S1, where the red line has a stronger slope towards low SPEI values (drought). Can you make a similar figure to S1, just for PET, to quantify this?

Line 225: your reference period for circulation is different from the period for SPI/SPEI calibration, is this done on purpose?

Line 262: “Whilst the aforementioned two variables are hypothesised to remain constant across time for each reanalysis, this is not the case for the former.” I’m not a native speaker, but I don’t think this makes grammatical sense. ‘The former and the latter’, refers to a list of two items, you have three.

Line 272: the Climate Explorer asks to be referred to as a dataset/tool as well, see https://climexp.knmi.nl/citation.cgi.

Figure 3: the SPI reference period is different in the caption than in the main text, the y-label should state SPI-3 [10-year], the figure doesn’t include labels a) and b), I suggest adding to the caption that the shaded area is in the colour of the dataset with the large error relative to the station measurement.

Line 357: why does ERA5 ‘support’ this finding, if it is the only reanalysis in Figure 5 that includes this decade?

Line 362: I’m not sure what you mean with 16th driest conditions here.

Figure 5: please do not use the rainbow colour scale, and please consider having a discrete colour bar (e.g. 10 colours) rather than a continuous, to help reading the figure. The discs are very difficult to see, maybe you can consider a different way of showing those. Just brainstorming: an additional subfigure with a line graph, or larger discs below the map.

Line 369: what is the mean dryness? If it is the mean SPEI3 value, is it not by definition that the two quantities are closely correlated?

Figure 6: a) please replace the rainbow colour scale, consider choosing a white-centred colour scale instead. Why is De Bilt/Frankfurt/Paris not included as a disc? b) I had to zoom in to about 200% to be able to read these graphs, consider modifying such that that is no issue anymore (line width, text size, font weight etc). Maybe all the slope values and p-values can be written in a table?

Line 380: this is slightly silly thing, but SPI or SPEI don’t measure ‘dryness’, they are an indicator of the relative dry or wet state of the past months. As such I don’t think the term ‘dryness trends’ are correct, maybe just (meteo)hydrological trends? Also in caption of section 7.3.

Figure 7: can you add somewhere (maybe in the caption) over how many seasons these were calculated? So for ERA5, 31% of how many droughts are related to the Baltic High? This may give the reader a sense of (statistical) stability.

Figure 8: a) please specify what the box (P25/P75?) and whiskers (min/max?) show, b) why not show this as four bar charts?, c) please replace the rainbow colour map, why not include De Bilt/Paris etc as a disc?

Line 510: “(Fig. 6b, Fig. S8b and Fig. S9b)” should this be Fig 8b?

Figure S11: What are the units on the y-axes? What is shown on the x-axis, are these all identified droughts consecutively? So a modified time axis?

Line 550: what do you mean here by “given the natural variability”? If 1921 was a large outlier, than would we expect an even larger (or slightly smaller if due to PET trends) to occur in the next 100 years?

Line 551: maybe add some literature on multi-year droughts in the region? Van der Wiel et al. (2023) or Van Mourik et al. (2025).

Line 565: “Some studies have been linking” should probably be “Some studies have linked”

Line 640: degree symbol missing. Furthermore, is CC scaling relevant to year-round precipitation? I always assumed it was mostly associated with convective summer rain events, rather than all precip year round? In the KNMI23 projections, extreme precip sometimes goes beyond CC (Van Dorland et al. 2023).

Line 735: I think you mean negative precipitation anomalies rather than precipitation deficits Deficits have been defined in many different ways in the literature so might cause confusion. (Also please check for other places in the manuscript.)

Line 746: Maybe add to this paragraph something short on temporally compounding or multi-year droughts?

References
Aalbers, Emma E., et al. "The 2018 west-central European drought projected in a warmer climate: how much drier can it get?." Natural Hazards and Earth System Sciences 23.5 (2023): 1921-1946.

Huang, Yafei, et al. "Long-term trends in agricultural droughts over Netherlands and Germany: how extreme was the year 2018?." Hydrology and Earth System Sciences Discussions 2021 (2021): 1-27.

Mens, Marjolein, et al. "Integrated drought risk assessment to support adaptive policy making in the Netherlands." Natural Hazards and Earth System Sciences Discussions 2022 (2022): 1-24.

Philip, Sjoukje Y., et al. "Regional differentiation in climate change induced drought trends in the Netherlands." Environmental Research Letters 15.9 (2020): 094081.

Van Dorland, R., et al. "KNMI national climate scenarios 2023 for the Netherlands." Report WR23-02, KNMI 605 (2023).

van Mourik, Jonna, et al. "Regional drivers and characteristics of multi-year droughts." Weather and Climate Extremes 48 (2025): 100748.

Van der Wiel, Karin, Geert Lenderink, and Hylke de Vries. "Physical storylines of future European drought events like 2018 based on ensemble climate modelling." Weather and Climate Extremes 33 (2021): 100350.

Van der Wiel, Karin, et al. "KNMI'23 climate scenarios for the Netherlands: storyline scenarios of regional climate change." Earth's Future 12.2 (2024): e2023EF003983.

Zscheischler, Jakob, and Erich M. Fischer. "The record-breaking compound hot and dry 2018 growing season in Germany." Weather and Climate Extremes 29 (2020): 100270.

Reply
Citation: https://doi.org/10.5194/egusphere-2025-5490-RC1

Emile Neimry, Hugues Goosse, and Mathieu Jonard

Supplement

https://doi.org/10.5194/egusphere-2025-5490-supplement

Emile Neimry, Hugues Goosse, and Mathieu Jonard

Viewed

Total article views: 392 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
203	169	20	392	58	28	23

HTML: 203
PDF: 169
XML: 20
Total: 392
Supplement: 58
BibTeX: 28
EndNote: 23

Views and downloads (calculated since 02 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	115	90	16	221
Jan 2026	84	79	4	167
Feb 2026	4	0	4

Cumulative views and downloads (calculated since 02 Dec 2025)

Month	HTML	PDF	XML	Total
Dec 2025	115	90	16	221
Jan 2026	84	79	4	167
Feb 2026	4	0	4

Viewed (geographical distribution)

Total article views: 371 (including HTML, PDF, and XML) Thereof 371 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 03 Feb 2026

Short summary

This study examines 180 years of drought variability in western Central Europe and its links to atmospheric circulation using three evaluated reanalysis datasets. Results reveal strong multidecadal variability, winter wetting, summer drying, and a growing influence of atmospheric evaporative demand. Four circulation patterns are identified, with recent droughts increasingly tied to the European High, marking a shift toward dynamics that intensify drought under climate change.


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