the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 07 Jul 2026
Development and comparison of documentary-instrumental and circulation-based multi-centennial precipitation reconstructions for Dublin, Ireland
Abstract. This study develops a multi-centennial reconstruction of precipitation for the Greater Dublin Area, Ireland, using documentary, instrumental, circulation-based, and proxy evidence. Three independent reconstructions are produced for two contrasting water-resource domains, including: a monthly scaled version of the Jenkinson documentary-instrumental rainfall dataset for 1711–1977 produced by an earlier analysis for the island of Ireland; a circulation-informed monthly statistical ensemble based on Lasso and Random Forest models for 1748–1994; and an annual oak cellulose δ¹⁸O reconstruction of May–August precipitation extending to 1200 CE. Reconstructions were assessed against 1 km gridded rainfall observations for the common overlap period 1865–1977 using correlation, bias, error metrics, SPI-based event detection, and agreement in ranked wet and dry extremes. Long-term UK rainfall series and an extended Dublin snow and sleet series were used as independent comparators. Both the Jenkinson and statistical ensemble reconstructions reproduce observed monthly, seasonal, and annual rainfall variability, but the Jenkinson reconstruction shows the strongest agreement across most metrics and better captures wet and dry extremes. The statistical ensemble provides an independent, physically interpretable reconstruction, although it tends to smooth variability and under-represent event magnitude. The δ¹⁸O reconstruction shows weaker interannual agreement with observed May–August rainfall but provides a valuable longer-term perspective on summer hydroclimatic variability. Divergence between the Jenkinson and ensemble reconstructions before the mid-nineteenth century, particularly in winter and spring, is consistent with possible snow- and sleet-related undercatch in early precipitation sources which affects the Jenkinson reconstruction. Summer rainfall during the eighteenth century remains uncertain, with evidence for increased regional hydroclimatic heterogeneity. Together, these reconstructions provide a stronger basis for contextualising historical droughts and pluvials affecting Dublin and for stress-testing water-resource systems beyond the instrumental period.
- Preprint
(4360 KB) - Metadata XML
-
Supplement
(354 KB) - BibTeX
- EndNote
Status: open (until 01 Sep 2026)
-
RC1: 'Comment on egusphere-2026-3352', Anonymous Referee #1, 22 Jul 2026
reply
-
AC1: 'Reply on RC1', Csaba Horvath, 04 Aug 2026
reply
Summary: The manuscript presents several reconstructions of monthly and seasonal precipitation in two regions close to Dublin, covering the past few centuries. The reconstructions are based on observational data sets from a gridded, station-based long record from the Irish Weather Service, documentary-observational data from two statistical models that take long records of circulation and temperature indices from the Western North Atlantic and Western Europe, and finally, a time series of Delta O18 that is linearly translated to seasonal (mostly summer) precipitation.
Thank you for this nice summary.
Recommendation: I have a very positive impression regarding the technical aspects of the manuscript, but I also have more critical suggestions regarding the target journal and the quality of the figures. Both critical remarks are related, and the way to address them depends on the objectives of the authors and the editor's opinion.
We thank the reviewer for their positive assessment of our manuscript. We address each point below.
Main points
1) The statistical and data-handling aspects are very good, and I wish to congratulate the authors for the thorough statistical analysis and the methodological descriptions. They are very clear and comprehensive. I do not have suggestions for any improvement in this regard, other than the minor point that the physical units of RMSE and MAE are missing in all tables.
Thank you, we address this point below.
2) My critical comments are as follows: whereas the statistical and data aspects are flawless, the manuscript does not contain any discussion of physical drivers for rainfall variability or of possible imprint of volcanic eruptions or greenhouse gases (trends in the 20th century, sudden changes during the Krakatoa or Tambora eruptions?, Any connection to the Year of the Slaughter in Ireland 1739-1740 (Jones and Briffa, Climatic Change (2006) 79: 361–379? Any connection to the Central England temperature series? The study is silent about all these aspects, which leads me to my second critical comment.
Thank you for these points. The aim of our paper is to present the reconstructions. The paper is more than just a data paper, which also relates to the next point. It details the development of methods for reconstructing and comparing reconstructions. These methods and findings (especially regarding sleet and snow) are of interest beyond the datasets produced. While it is beyond the scope of this paper to fully explore drivers and volcanic eruptions, we can provide insight into how known extreme years/periods are represented in the reconstructions, e.g., the “Year of the Slaughter”, 1816, etc. We provide some initial analysis at the end of the response document, which can form an additional results section.
3) Is Climate of the Past the correct target journal? Evaluating the present version of the manuscript, I would rather believe that it would be better hosted by ESSD, given that it is very data-centric and void of other climatological or dynamical elaborations. Depending on how this question is answered, the manuscript could go to ESSD with, in my opinion, rather minor revisions, or it may remain in Climate of the Past, but then a whole new section should be included, for example, focusing on the dynamical aspects that I mention in my point 2.
I do think that the reconstructions are very interesting, and they are definitely worth deeper exploration. I would welcome it if the authors wished to pursue the second path and expand the study. If not, however, I believe that the present version is a bit short of content for Climate of the Past. Alternatively, the authors may consider converting this manuscript to Part I of a two-part manuscript, with Part1 published in ESSD and Part 2 in Climate of the Past.
We thank the reviewer for this reflection and for highlighting how interesting the reconstructions are. However, we believe that Climate of the Past is the best home for this paper. The journal frequently publishes similar papers. For example, Herzschuh et al. (2026) provide quantitative climate reconstructions from sedimentary ancient DNA, and focus on the framework and applications of these reconstructions. Murphy et al. (2018) provide an equivalent rainfall reconstruction for the island of Ireland back to 1711. To appease the reviewer’s concern we propose adding an additional results section on trends in the reconstructions and a reflection on notable extremes.
4) Related to the previous point is the issue of the quality of the figures. They are all of low resolution, which in principle should not be a terrible problem. In this case, however, it does become an issue. I have been trying to discern by eye whether one can see an impact of volcanic eruptions in the series, or whether rainfall anomalies can be related to previously published temperature reconstructions, but the low resolution does not allow me to really assess the timing of peaks in the time series. To be fair, the data have been made available, but an interested reader should be able to clearly see all details of the time series in the manuscript. As all plots include several time series, with different colours that are not well defined in the low-resolution figures either, this is not easy at all. The clear value of the reconstructions is somewhat diminished by the low figure resolution.
Apologies for this, not sure why. We will ensure all figures are high resolution.
Minor points
5) Figure 1, please include geographical coordinates.
Will do this.
6) ‘We assess the consistency of the top 20 wettest and driest years between 1865 and 1977 across each reconstruction and observations using the Jaccard Similarity Coefficient...’
This suggestion is more a matter of taste but it would help the reader to add that the JI is related to the more usual F1-score by JI=F1/(2-F1). Or is JI in this case based on three (plus,neutral,minus) categories?
Thanks, we can add that detail.
7) Table 2 ,3 4,5 , include units of RMSE, MAE and PBIAS
Thanks, will do this
8) All reconstructions seem to underestimate the observed variability. For instance, in the scatter plots shown in Figures 4 and 5, the range of the x-axis (observations) is always larger than the y-axis (reconstructions). This difference in range can be misleading. The text does discuss the underestimation of extremes, but I do suggest redrawing the scatter plots using the same range for both axes.
Related to this point, it seems to me, just by eyeballing, that these scatter plots indicate a non-linear link between observations and reconstructions. Even with the unequal axis ranges, it seems that the data cloud is ‘bent’ . This, of course, may be related to the underestimation of extremes, or perhaps a deeper issue throughout the whole range of the reconstructions.
Thanks, we can redraw the scatter plots to have the same axis ranges and insert a 1:1 line to more clearly show this. We disagree that eyeballing indicates non-linear behaviour. Statistical evaluation shows no evidence of systematic non-linear behaviour across the reconstructions, most likely due to underestimation of high rainfall totals. We also note that the primary use of these reconstructions will be for historical drought analysis in the Dublin region, which is a focus of ongoing work.
References
Herzschuh, U., Böhmer, T., Jia, W., and Lisovski, S.: Quantitative climate reconstruction from sedimentary ancient DNA: framework, validation and application, Clim. Past, 22, 1159–1180, https://doi.org/10.5194/cp-22-1159-2026, 2026.
Murphy, C., Broderick, C., Burt, T. P., Curley, M., Duffy, C., Hall, J., Harrigan, S., Matthews, T. K. R., Macdonald, N., McCarthy, G., McCarthy, M. P., Mullan, D., Noone, S., Osborn, T. J., Ryan, C., Sweeney, J., Thorne, P. W., Walsh, S., and Wilby, R. L.: A 305-year continuous monthly rainfall series for the island of Ireland (1711–2016), Clim. Past, 14, 413–440, https://doi.org/10.5194/cp-14-413-2018, 2018.
Citation: https://doi.org/10.5194/egusphere-2026-3352-AC1
-
AC1: 'Reply on RC1', Csaba Horvath, 04 Aug 2026
reply
-
RC2: 'Comment on egusphere-2026-3352', Anonymous Referee #2, 07 Aug 2026
reply
In the temperate climate zone, precipitation is the meteorological variable characterised by the greatest temporal and spatial variability. It is also one of the most important meteorological variables, second only to air temperature. Therefore, reconstructing precipitation patterns over the longest possible period, particularly over the last millennium, is both important and scientifically valuable.
In the manuscript under review, the authors undertook this task for Dublin and the surrounding area. They presented three reconstructions of precipitation totals that differ in their estimated values, with each providing specific insights into the long-term variability of precipitation. The manuscript focuses primarily on the methodology used to reconstruct precipitation series from both instrumental and proxy data.
Each reconstructed precipitation series was calibrated and verified against instrumental observations using appropriate statistical methods. In conclusion, I believe that the authors have demonstrated in considerable detail that their reconstructed series are reliable and suitable for a variety of applications.
Main points
- The main weakness of the manuscript is the limited scope of the climatological analysis. The authors provide only lists of the 20 wettest and 20 driest years. It would also be valuable to identify the wettest and driest multi-year periods and to attempt to explain their possible causes. Beyond the statistical reconstruction itself, the authors could investigate the atmospheric circulation patterns or other factors responsible for these periods.
- Since several UK precipitation datasets were used in the reconstruction, the manuscript would also benefit from a brief comparison of precipitation conditions in the UK and the Dublin area, with particular emphasis on their similarities and differences. Simply stating that circulation patterns are similar in both regions is not entirely sufficient, as substantial differences may be expected, particularly in summer, as is also the case in continental Europe.
- The area referred to as “SA upland” should also be described more precisely in terms of elevation range, topography, and exposure to incoming air masses, all of which may significantly affect precipitation totals.
- Given the manuscript’s predominantly technical focus, particularly its extensive presentation of the statistical methods used to calibrate and verify the reconstructed precipitation series and to quantify their uncertainty, it may be better suited to a data-oriented journal such as Earth System Science Data or Scientific Data. One possible solution would be to move a substantial proportion of the statistical analyses to the Supplementary Material and to place greater emphasis in the main text on the climatological interpretation of the reconstructed precipitation series, including long-term variability, trends, and related aspects.
Minor points
Line 41 – Replace “climatic variability” with the more natural expression “natural climate variability”.
Line 44 – Add the missing full stop after “(Murphy, 2019)”.
Lines 115–117 – I suggest providing information on the elevation range and topography of the study area.
Figure 1 – The colour used for the urban area differs between the legend and the map; please correct this inconsistency. I also suggest using a topographic map that shows the relief and elevation. Please add geographical coordinates as well.
Line 134 – It would be useful to provide information on the number of precipitation stations located within the GDA, their exact locations, and their correspondence with the nearest grid cells in the dataset. The relevant grid cells should also be shown on the map.
Line 160 – Line 160 – Add a full stop after “al”: “et al” should be changed to “et al.”
Lines 169–175 – It is unclear which specific data from the series described above, and for which corresponding periods, were used to construct the final precipitation series. When referring to the late twentieth century, please specify the exact year.
Line 196 – Add a space: “to1994” should be changed to “to 1994”.
Table 1 – Please add information on the time period covered by each predictor.
Line 219 – Delete “May–August”, as it is repeated.
Lines 245–249 – The grid cells should be shown in Figure 1.
Figures 6 and 7 – Difficult to read. I suggest also showing, for a common period, the differences between the statistical and Jenkinson reconstructions and the observations.
Figure 8 – Too many variables are presented in a single figure, making the differences difficult to discern.
Figure 12 – There are far too many lines or curves in a single figure, making it largely illegible.
Figure 13 – The figure should show the differences relative to the observations more clearly. In its current form, it is difficult to read.
Citation: https://doi.org/10.5194/egusphere-2026-3352-RC2
Data sets
Multi-centennial Rainfall Reconstructions for the Greater Dublin Area, Ireland Version: 1.0 C. Horvath et al. https://doi.org/10.5281/zenodo.20605304
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 84 | 31 | 13 | 128 | 19 | 12 | 13 |
- HTML: 84
- PDF: 31
- XML: 13
- Total: 128
- Supplement: 19
- BibTeX: 12
- EndNote: 13
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Summary: The manuscript presents several reconstructions of monthly and seasonal precipitation in two regions close to Dublin, covering the past few centuries. The reconstructions are based on observational data sets from a gridded, station-based long record from the Irish Weather Service, documentary-observational data from two statistical models that take long records of circulation and temperature indices from the Western North Atlantic and Western Europe, and finally, a time series of Delta O18 that is linearly translated to seasonal (mostly summer) precipitation.
Recommendation: I have a very positive impression regarding the technical aspects of the manuscript, but I also have more critical suggestions regarding the target journal and the quality of the figures. Both critical remarks are related, and the way to address them depends on the objectives of the authors and the editor's opinion.
Main points
1) The statistical and data-handling aspects are very good, and I wish to congratulate the authors for the thorough statistical analysis and the methodological descriptions. They are very clear and comprehensive. I do not have suggestions for any improvement in this regard, other than the minor point that the physical units of RMSE and MAE are missing in all tables.
2) My critical comments are as follows: whereas the statistical and data aspects are flawless, the manuscript does not contain any discussion of physical drivers for rainfall variability or of possible imprint of volcanic eruptions or greenhouse gases (trends in the 20th century, sudden changes during the Krakatoa or Tambora eruptions?, Any connection to the Year of the Slaughter in Ireland 1739-1740 (Jones and Briffa, Climatic Change (2006) 79: 361–379? Any connection to the Central England temperature series? The study is silent about all these aspects, which leads me to my second critical comment.
3) Is Climate of the Past the correct target journal? Evaluating the present version of the manuscript, I would rather believe that it would be better hosted by ESSD, given that it is very data-centric and void of other climatological or dynamical elaborations. Depending on how this question is answered, the manuscript could go to ESSD with, in my opinion, rather minor revisions, or it may remain in Climate of the Past, but then a whole new section should be included, for example, focusing on the dynamical aspects that I mention in my point 2.
I do think that the reconstructions are very interesting, and they are definitely worth deeper exploration. I would welcome it if the authors wished to pursue the second path and expand the study. If not, however, I believe that the present version is a bit short of content for Climate of the Past. Alternatively, the authors may consider converting this manuscript to Part I of a two-part manuscript, with Part1 published in ESSD and Part 2 in Climate of the Past.
4) Related to the previous point is the issue of the quality of the figures. They are all of low resolution, which in principle should not be a terrible problem. In this case, however, it does become an issue. I have been trying to discern by eye whether one can see an impact of volcanic eruptions in the series, or whether rainfall anomalies can be related to previously published temperature reconstructions, but the low resolution does not allow me to really assess the timing of peaks in the time series. To be fair, the data have been made available, but an interested reader should be able to clearly see all details of the time series in the manuscript. As all plots include several time series, with different colours that are not well defined in the low-resolution figures either, this is not easy at all. The clear value of the reconstructions is somewhat diminished by the low figure resolution.
Minor points
5) Figure 1, please include geographical coordinates.
6) ‘We assess the consistency of the top 20 wettest and driest years between 1865 and 1977 across
each reconstruction and observations using the Jaccard Similarity Coefficient...’
This suggestion is more a matter of taste but it would help the reader to add that the JI is related to the more usual F1-score by JI=F1/(2-F1). Or is JI in this case based on three (plus,neutral,minus) categories?
7) Table 2 ,3 4,5 , include units of RMSE, MAE and PBIAS
8) All reconstructions seem to underestimate the observed variability. For instance, in the scatter plots shown in Figures 4 and 5, the range of the x-axis (observations) is always larger than the y-axis (reconstructions). This difference in range can be misleading. The text does discuss the underestimation of extremes, but I do suggest redrawing the scatter plots using the same range for both axes.
Related to this point, it seems to me, just by eyeballing, that these scatter plots indicate a non-linear link between observations and reconstructions. Even with the unequal axis ranges, it seems that the data cloud is ‘bent’ . This, of course, may be related to the underestimation of extremes, or perhaps a deeper issue throughout the whole range of the reconstructions.